Review of deep learning: concepts, CNN architectures, challenges, applications, future directions

Laith Alzubaidi, Jinglan Zhang, Amjad J. Humaidi, Ayad Q. Al-Dujaili, Ye Duan, Omran Al-Shamma, José Santamaría, Mohammed A. Fadhel, Muthana Al‐Amidie, Laith Farhan · Journal Of Big Data · 2021 · artificial-intelligence · 10.1186/s40537-021-00444-8 · record alzubaidi2021-review-of-deep-learning-concepts-c
tag layer · index · concepts · authors · עברית

Where this record's text came from. 2026-08-21 and 2026-08-28 schema-0.3 passes: every quote verified as an exact substring of the staged text and anchored at merge time; concept terms verified present; every edge endpoint an existing node.

This record’s own method is review, and the tags below follow from that. A review’s findings, quantities and statistics are things stated IN this paper, and some of them report results obtained elsewhere. No tag on this page names a different author. Measured across every review in this corpus, 81% of a review’s tags carry no citation at all — most of them are the review speaking. The corpus does not adjudicate the rest: read an unmarked tag as a claim this paper makes, not necessarily one it made first.

How to read this page. The article itself is not reproduced here — this paper is not cleared for redistribution, so what follows is the tag layer alone. The evidence sentences are withheld with the article. The table at the foot lists every tag the record carries; each row names the facet it belongs to, and the whose words column says whether the tag records the paper's own wording or the tagger's sentence about the paper. Every tag links to each other paper in the corpus carrying it. The machine-readable edition of the same thing is the TEI file beside this one, where the tags live in standOff, and the article is at its publisher via the link below. When we have changed our mind. Some entries in the table begin with a word in capitals and the reason for it. EVIDENCE CORRECTED — the finding still stands, but we had quoted the wrong sentence from the paper, and the quotation has been replaced with the right one. RETRACTED — the entry itself was wrong; it stays on the page, marked, with the reason, rather than being deleted. WITHDRAWN AND REPLACED — the entry was wrong, and a corrected version of it appears in the entries below. Each mark carries its own reason, in brackets, because the reasons differ from one entry to the next. An entry we have WITHDRAWN is removed from this page and kept in the record, with its reason. A line reading Audit and a date is a later re-reading of the entry against its whole sentence: DOUBTFUL means it was kept, and the reason says what is loose about it. Quotation marks inside the tagger's sentence. Where a tag is the tagger's own sentence, a phrase in quotation marks inside it is borrowed from the paper — the tagger's wording around it is not. The paper's own sentence is the evidence quote for that row; on this page it is withheld with the article, and the article is at its publisher. Nothing inside quotation marks anywhere on this page is ours: if it is quoted, it is the paper's. Whose words a concept carries. A concept entry records a term the paper itself uses, so the term, the sense given for it, and its evidence are the paper's own words, quoted where the paper fixes them. This is the opposite of a sense entry in the tag table, which is the tagger's sentence about how the paper uses a term. Where a concept entry could not be grounded in the paper's words, the entry says so on its own row.

Other fields define these differently the same word, defined another way in another discipline
Precision 10 fields · Accuracy 9 fields · Sensitivity 8 fields · Specificity 8 fields · overfitting 6 fields · ML 5 fields · and 16 more →

Tags on this paper

Every tag the record carries, each linking to every other paper in the corpus that carries it. A tag appears here whether or not its words occur in the article — the spans below are only the ones a sentence in this paper licenses. Of the 171 concept tags on this paper, all 171 are grounded in a quoted sentence. All carry the tagger's gloss of what the term means in this paper, which is not something the article can license.

representation learning RL deep reinforcement learning DL ML TL Same-domain TL underspecification stress tests Sparse Connectivity Weight Sharing Dropout Dying ReLU catastrophic forgetting vanishing gradient problem overfitting black box imbalanced hyper-parameter selection lightweight models trained from scratch feature engineering histogram of oriented gradients (HOG) scale-invariant feature transform (SIFT) bag of words (BoW) feature extraction multi-layer data representation architecture low-level features high-level features unsupervised partially supervised (semi-supervised) supervised labeled data recurrent neural networks (RNNs) convolutional neural networks (CNNs) deep neural networks (DNNs) gated recurrent units (GRUs) long short-term memory (LSTM) approaches decision boundary generative networks dimensionality reduction clustering restricted Boltzmann machines auto-encoders GANs reward function environment supervised learning RNN sequential data short-term memory state (hidden) layer exploding gradient LSTM memory blocks memory cells gated units conventional networks CNN equivalent representations sparse interactions parameter sharing shared weights local connections 2D input-data structures convolution layers sub-sampling (pooling) layers FC layers feature maps softmax support vector machines (SVMs) kernel weight convolutional operation feature map stride padding convolutional layer convolutional filters pooling layer pooling methods max pooling min pooling global average pooling (GAP) global max pooling activation function Sigmoid Tanh ReLU Leaky ReLU Noisy ReLU Parametric Linear Units Fully Connected Layer loss functions Cross-Entropy or Softmax Loss Function log loss function Euclidean Loss Function mean square error Hinge Loss Function Drop-Weights Data Augmentation Batch Normalization internal covariance shift over-fitted under-fitted model just-fitted learning rate training epoch Batch Gradient Descent Stochastic Gradient Descent Mini-batch Gradient Descent Adaptive Moment Estimation (Adam) Adam learning strategy residual links deep network convergence highway network ResNet DenseNet CapsuleNet CapsNet High-Resolution Network (HRNet) HRNet training data simulated data data augmentation techniques data-space solutions model interpretability perturbation-based approaches back-propagation-based techniques score of confidence temperature scaling incorporating new information into a plain DL model parameter pruning knowledge distillation compact convolution filters low-rank factorization lack of training data classification localization detection segmentation registration Target Selection Feature Matching Pose Optimization CPU GPU FPGA Accuracy Sensitivity or Recall Specificity Precision F1-Score J Score Youdens J statistic Area Under the ROC Curve RNN category increasing the stride value a feature map of lower dimensions sub-sampling layers a reduction in the network parameters how confident the model is in its prediction overfitting problem visual cortex exploding gradient and vanishing problems model compression DNN training inference accelerators Sensitivity Recall vanishing gradient through repeated multiplication catastrophic forgetting by fine-tuning domain mismatch defeating transfer underspecification selecting among equivalent solutions AUC amount of data review

Method, as described: a literature survey with a declared search framework - named keyword strings, named publishers, and a stated year distribution over 300+ papers

Legend

relation to other work	`claim_relations`	how this paper stands to another: agrees, contradicts, extends, qualifies
claim type	`claim_type`	the shape of the claim: existence, mechanism, method, critique, prediction, null
concept edge	`concept_edges`	an asserted relation between two concepts, with the sentence asserting it
concept	`concepts`	an idea the paper works with, quoted where the paper uses it
content tag	`content_tags`	a finding, definition, quantity, statistic or variable named in the paper
contribution	`contribution`	what the paper claims to add
control	`controls`	an experiment run to eliminate a rival explanation
limitation	`limitations`	what the paper says its evidence cannot support
mechanism	`mechanisms`	how the reported effect is said to come about
method	`method`	how the paper produced its evidence
object of study	`object_of_study`	what the paper is about — the thing it investigates
phenomenon	`phenomenon`	the effect or regularity under study
research question	`research_question`	the question the paper sets itself
sense	`senses`	a distinct use of a term, quoted where the paper fixes it
transfer	`transfers`	a method or concept carried in from another field
unrecordable relation	`unhoused_relations`	a relation the paper asserts that this vocabulary cannot express, with the reason

The paper

The full text is not reproduced here. This article is not cleared for redistribution, so the page carries the tag layer only. Its evidence sentences are withheld; the grounding status remains with each tag, and the article itself is at https://doi.org/10.1186/s40537-021-00444-8. The reading view exists for the papers whose licence permits it.

Every tag on this page

claim_relations 3 · claim_type 1 · concept_edges 22 · concepts 171 · content_tags 20 · contribution 1 · controls 3 · limitations 6 · mechanisms 4 · method 1 · object_of_study 1 · phenomenon 1 · research_question 1 · senses 16 · transfers 2 · unhoused_relations 10

Relations on this page carry a quotation for the claim. Where a relation was separately evidenced — and so tested against the vocabulary for its type — the row says so; 22 of 22 relations here carry that mark. The rest are the tagger’s reading of the sentence quoted beside them.

facet	tag	field	conf.	whose words
object_of_study	deep learning as a field - its concepts, network types, CNN architectures, named challenges, applications and computational substrate Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`object_of_study`		the paper's words
research_question	whether one review can supply a full understanding of DL, against a review literature in which each paper covers a single aspect Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`research_question`		the paper's words
contribution	a holistic survey of DL with twelve named challenges and their proposed solutions, a CNN-architecture lineage from AlexNet to HRNet, and a declared search methodology Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`contribution`		the paper's words
claim_type	critique Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_type`		the paper's words
phenomenon	a field whose knowledge is fragmented across partial reviews, so that assembling a working understanding requires reading past the reviews Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`phenomenon`		the paper's words
method	review Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: review	`method`		the paper's words
mechanisms	back-propagation multiplies each layer's derivative down the network, so with a squashing activation like the sigmoid, N small derivatives multiplied together make the gradient decline exponentially - and the first layers, which recognise the essential elements of the input, are the ones that stop updating Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: vanishing gradient through repeated multiplication	`mechanisms[0]`	0.95	the tagger's words
mechanisms	incorporating new information into a trained model interferes with what it already learned; fine-tuning on only the new class leaves performance on the old classes unsuccessful, so continually renewed data forces retraining from scratch Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: catastrophic forgetting by fine-tuning	`mechanisms[1]`	0.95	the tagger's words
mechanisms	pre-training transfers representations learned from the source distribution; where the source is natural images and the target is raw medical images the learned representation does not apply, so the transferred model performs no better than one trained from scratch Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: domain mismatch defeating transfer	`mechanisms[2]`	0.95	the tagger's words
mechanisms	many parameter settings satisfy the training objective equally well, so the pipeline does not determine which one is chosen; small modifications force the model toward a completely different solution and hence different predictions in deployment Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: underspecification selecting among equivalent solutions	`mechanisms[3]`	0.95	the tagger's words
limitations	the paper recommends transfer learning for scarce data and then reports that cross-domain TL does not significantly affect medical imaging performance - its own central recommendation carries a negative result three paragraphs later Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[0]`	0.95	the tagger's words
limitations	`RL` is defined as representation learning in the introduction and used for reinforcement learning in the classification section; a reader carrying the first definition forward misreads the second Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[1]`	0.95	the tagger's words
limitations	underspecification puts major constraints on the credibility of ML predictions and the proposed remedy - stress tests - requires already knowing how the model can fail Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[2]`	0.95	the tagger's words
limitations	CNN performance is greatly influenced by hyper-parameter selection, so any small change affects the result - which limits how much a reported architecture comparison can be attributed to architecture Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[3]`	0.95	the tagger's words
limitations	labelling remains the binding constraint: crowdsourced workers cannot annotate medical or biological images, expert labelling is costly and slow, so producing the volume of labels deep networks need is called unfeasible Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[4]`	0.95	the tagger's words
limitations	despite the stated transition toward semi-supervised and unsupervised learning, most current deep-learning models still use supervised learning - so the data constraint above is not being relieved Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`limitations[5]`	0.95	the tagger's words
transfers	the architecture of the mammalian visual cortex, carried into network design - the CNN's local receptive fields are presented as a simulation of cells that sense only small regions of a scene Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`transfers[0]`	0.95	the tagger's words
transfers	pre-trained representations carried from natural-image datasets into medical imaging, and the paper's own report that this particular transfer does not work Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`transfers[1]`	0.95	the tagger's words
controls	a declared corpus for the review - a stated floor on papers read and their distribution by year - so that 'we focused on recent work' is a number rather than a claim Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. eliminates: a review's coverage claim resting on the reviewers' sense of having read enough · outcome: held	`controls[0]`	0.95	the tagger's words
controls	declared review objectives - list and define approaches, list architectures, present challenges, assess applications and computational approaches - fixed in advance of the reading Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. eliminates: a review whose scope is settled after the reading, by fixing the five objectives in advance · outcome (stated by the tagger, no quotation): held	`controls[1]`	0.85	the tagger's words
controls	cross-domain transfer learning checked against models trained from scratch - the control that decides whether transfer contributed anything, run by others and reported here Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. eliminates: that the reported performance is owed to the pre-trained source domain rather than to the architecture and the target data - models trained from scratch are the comparison that decides it · outcome: absent-stated-about-other	`controls[2]`	0.95	the tagger's words
concepts	representation learning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: representation learning	`concepts[0]`	0.95	the paper's words
concepts	RL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: RL	`concepts[1]`	0.95	the paper's words
concepts	deep reinforcement learning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: deep reinforcement learning	`concepts[2]`	0.95	the paper's words
concepts	DL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: DL	`concepts[3]`	0.95	the paper's words
concepts	ML Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: ML	`concepts[4]`	0.95	the paper's words
concepts	TL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: TL	`concepts[5]`	0.95	the paper's words
concepts	Same-domain TL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Same-domain TL	`concepts[6]`	0.95	the paper's words
concepts	underspecification Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: underspecification	`concepts[7]`	0.95	the paper's words
concepts	stress tests Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: stress tests	`concepts[8]`	0.95	the paper's words
concepts	Sparse Connectivity Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sparse Connectivity	`concepts[9]`	0.95	the paper's words
concepts	Weight Sharing Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Weight Sharing	`concepts[10]`	0.95	the paper's words
concepts	Dropout Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Dropout	`concepts[11]`	0.95	the paper's words
concepts	Dying ReLU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Dying ReLU	`concepts[12]`	0.95	the paper's words
concepts	catastrophic forgetting Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: catastrophic forgetting	`concepts[13]`	0.85	the paper's words
concepts	vanishing gradient problem Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: vanishing gradient problem	`concepts[14]`	0.95	the paper's words
concepts	overfitting Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: overfitting	`concepts[15]`	0.95	the paper's words
concepts	black box Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: black box	`concepts[16]`	0.95	the paper's words
concepts	imbalanced Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: imbalanced	`concepts[17]`	0.95	the paper's words
concepts	hyper-parameter selection Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: hyper-parameter selection	`concepts[18]`	0.95	the paper's words
concepts	lightweight models trained from scratch Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: lightweight models trained from scratch	`concepts[19]`	0.95	the paper's words
concepts	feature engineering Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: feature engineering	`concepts[20]`	0.95	the paper's words
concepts	histogram of oriented gradients (HOG) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: histogram of oriented gradients (HOG)	`concepts[21]`	0.95	the paper's words
concepts	scale-invariant feature transform (SIFT) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: scale-invariant feature transform (SIFT)	`concepts[22]`	0.95	the paper's words
concepts	bag of words (BoW) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: bag of words (BoW)	`concepts[23]`	0.95	the paper's words
concepts	feature extraction Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: feature extraction	`concepts[24]`	0.95	the paper's words
concepts	multi-layer data representation architecture Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: multi-layer data representation architecture	`concepts[25]`	0.95	the paper's words
concepts	low-level features Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: low-level features	`concepts[26]`	0.95	the paper's words
concepts	high-level features Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: high-level features	`concepts[27]`	0.95	the paper's words
concepts	unsupervised Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: unsupervised	`concepts[28]`	0.95	the paper's words
concepts	partially supervised (semi-supervised) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: partially supervised (semi-supervised)	`concepts[29]`	0.95	the paper's words
concepts	supervised Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: supervised	`concepts[30]`	0.95	the paper's words
concepts	labeled data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: labeled data	`concepts[31]`	0.95	the paper's words
concepts	recurrent neural networks (RNNs) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: recurrent neural networks (RNNs)	`concepts[32]`	0.95	the paper's words
concepts	convolutional neural networks (CNNs) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: convolutional neural networks (CNNs)	`concepts[33]`	0.95	the paper's words
concepts	deep neural networks (DNNs) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: deep neural networks (DNNs)	`concepts[34]`	0.95	the paper's words
concepts	gated recurrent units (GRUs) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: gated recurrent units (GRUs)	`concepts[35]`	0.95	the paper's words
concepts	long short-term memory (LSTM) approaches Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: long short-term memory (LSTM) approaches	`concepts[36]`	0.95	the paper's words
concepts	decision boundary Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: decision boundary	`concepts[37]`	0.95	the paper's words
concepts	generative networks Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: generative networks	`concepts[38]`	0.95	the paper's words
concepts	dimensionality reduction Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: dimensionality reduction	`concepts[39]`	0.95	the paper's words
concepts	clustering Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: clustering	`concepts[40]`	0.95	the paper's words
concepts	restricted Boltzmann machines Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: restricted Boltzmann machines	`concepts[41]`	0.95	the paper's words
concepts	auto-encoders Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: auto-encoders	`concepts[42]`	0.95	the paper's words
concepts	GANs Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: GANs	`concepts[43]`	0.95	the paper's words
concepts	reward function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: reward function	`concepts[44]`	0.95	the paper's words
concepts	environment Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: environment	`concepts[45]`	0.95	the paper's words
concepts	supervised learning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: supervised learning	`concepts[46]`	0.95	the paper's words
concepts	RNN Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: RNN	`concepts[47]`	0.95	the paper's words
concepts	sequential data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: sequential data	`concepts[48]`	0.95	the paper's words
concepts	short-term memory Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: short-term memory	`concepts[49]`	0.95	the paper's words
concepts	state (hidden) layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: state (hidden) layer	`concepts[50]`	0.95	the paper's words
concepts	exploding gradient Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: exploding gradient	`concepts[51]`	0.95	the paper's words
concepts	LSTM Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: LSTM	`concepts[52]`	0.95	the paper's words
concepts	memory blocks Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: memory blocks	`concepts[53]`	0.95	the paper's words
concepts	memory cells Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: memory cells	`concepts[54]`	0.95	the paper's words
concepts	gated units Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: gated units	`concepts[55]`	0.95	the paper's words
concepts	conventional networks Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: conventional networks	`concepts[56]`	0.95	the paper's words
concepts	CNN Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: CNN	`concepts[57]`	0.95	the paper's words
concepts	equivalent representations Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: equivalent representations	`concepts[58]`	0.95	the paper's words
concepts	sparse interactions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: sparse interactions	`concepts[59]`	0.95	the paper's words
concepts	parameter sharing Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: parameter sharing	`concepts[60]`	0.95	the paper's words
concepts	shared weights Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: shared weights	`concepts[61]`	0.95	the paper's words
concepts	local connections Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: local connections	`concepts[62]`	0.95	the paper's words
concepts	2D input-data structures Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: 2D input-data structures	`concepts[63]`	0.95	the paper's words
concepts	convolution layers Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: convolution layers	`concepts[64]`	0.95	the paper's words
concepts	sub-sampling (pooling) layers Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: sub-sampling (pooling) layers	`concepts[65]`	0.95	the paper's words
concepts	FC layers Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: FC layers	`concepts[66]`	0.95	the paper's words
concepts	feature maps Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: feature maps	`concepts[67]`	0.95	the paper's words
concepts	softmax Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: softmax	`concepts[68]`	0.95	the paper's words
concepts	support vector machines (SVMs) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: support vector machines (SVMs)	`concepts[69]`	0.95	the paper's words
concepts	kernel weight Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: kernel weight	`concepts[70]`	0.95	the paper's words
concepts	convolutional operation Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: convolutional operation	`concepts[71]`	0.95	the paper's words
concepts	feature map Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: feature map	`concepts[72]`	0.95	the paper's words
concepts	stride Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: stride	`concepts[73]`	0.95	the paper's words
concepts	padding Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: padding	`concepts[74]`	0.95	the paper's words
concepts	convolutional layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: convolutional layer	`concepts[75]`	0.95	the paper's words
concepts	convolutional filters Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: convolutional filters	`concepts[76]`	0.95	the paper's words
concepts	pooling layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: pooling layer	`concepts[77]`	0.95	the paper's words
concepts	pooling methods Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: pooling methods	`concepts[78]`	0.95	the paper's words
concepts	max pooling Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: max pooling	`concepts[79]`	0.95	the paper's words
concepts	min pooling Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: min pooling	`concepts[80]`	0.95	the paper's words
concepts	global average pooling (GAP) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: global average pooling (GAP)	`concepts[81]`	0.95	the paper's words
concepts	global max pooling Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: global max pooling	`concepts[82]`	0.95	the paper's words
concepts	activation function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: activation function	`concepts[83]`	0.95	the paper's words
concepts	Sigmoid Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sigmoid	`concepts[84]`	0.95	the paper's words
concepts	Tanh Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Tanh	`concepts[85]`	0.95	the paper's words
concepts	ReLU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: ReLU	`concepts[86]`	0.95	the paper's words
concepts	Leaky ReLU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Leaky ReLU	`concepts[87]`	0.95	the paper's words
concepts	Noisy ReLU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Noisy ReLU	`concepts[88]`	0.95	the paper's words
concepts	Parametric Linear Units Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Parametric Linear Units	`concepts[89]`	0.95	the paper's words
concepts	Fully Connected Layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Fully Connected Layer	`concepts[90]`	0.95	the paper's words
concepts	loss functions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: loss functions	`concepts[91]`	0.95	the paper's words
concepts	Cross-Entropy or Softmax Loss Function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Cross-Entropy or Softmax Loss Function	`concepts[92]`	0.95	the paper's words
concepts	log loss function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: log loss function	`concepts[93]`	0.95	the paper's words
concepts	Euclidean Loss Function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Euclidean Loss Function	`concepts[94]`	0.95	the paper's words
concepts	mean square error Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: mean square error	`concepts[95]`	0.95	the paper's words
concepts	Hinge Loss Function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Hinge Loss Function	`concepts[96]`	0.95	the paper's words
concepts	Drop-Weights Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Drop-Weights	`concepts[97]`	0.95	the paper's words
concepts	Data Augmentation Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Data Augmentation	`concepts[98]`	0.95	the paper's words
concepts	Batch Normalization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Batch Normalization	`concepts[99]`	0.95	the paper's words
concepts	internal covariance shift Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: internal covariance shift	`concepts[100]`	0.95	the paper's words
concepts	over-fitted Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: over-fitted	`concepts[101]`	0.95	the paper's words
concepts	under-fitted model Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: under-fitted model	`concepts[102]`	0.95	the paper's words
concepts	just-fitted Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: just-fitted	`concepts[103]`	0.95	the paper's words
concepts	learning rate Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: learning rate	`concepts[104]`	0.95	the paper's words
concepts	training epoch Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: training epoch	`concepts[105]`	0.95	the paper's words
concepts	Batch Gradient Descent Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Batch Gradient Descent	`concepts[106]`	0.95	the paper's words
concepts	Stochastic Gradient Descent Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Stochastic Gradient Descent	`concepts[107]`	0.95	the paper's words
concepts	Mini-batch Gradient Descent Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Mini-batch Gradient Descent	`concepts[108]`	0.95	the paper's words
concepts	Adaptive Moment Estimation (Adam) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Adaptive Moment Estimation (Adam)	`concepts[109]`	0.95	the paper's words
concepts	Adam Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Adam	`concepts[110]`	0.95	the paper's words
concepts	learning strategy Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: learning strategy	`concepts[111]`	0.95	the paper's words
concepts	residual links Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: residual links	`concepts[112]`	0.95	the paper's words
concepts	deep network convergence Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: deep network convergence	`concepts[113]`	0.95	the paper's words
concepts	highway network Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: highway network	`concepts[114]`	0.95	the paper's words
concepts	ResNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: ResNet	`concepts[115]`	0.95	the paper's words
concepts	DenseNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: DenseNet	`concepts[116]`	0.95	the paper's words
concepts	CapsuleNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: CapsuleNet	`concepts[117]`	0.95	the paper's words
concepts	CapsNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: CapsNet	`concepts[118]`	0.95	the paper's words
concepts	High-Resolution Network (HRNet) Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: High-Resolution Network (HRNet)	`concepts[119]`	0.95	the paper's words
concepts	HRNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: HRNet	`concepts[120]`	0.95	the paper's words
concepts	training data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: training data	`concepts[121]`	0.95	the paper's words
concepts	simulated data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: simulated data	`concepts[122]`	0.95	the paper's words
concepts	data augmentation techniques Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: data augmentation techniques	`concepts[123]`	0.95	the paper's words
concepts	data-space solutions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: data-space solutions	`concepts[124]`	0.95	the paper's words
concepts	model interpretability Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: model interpretability	`concepts[125]`	0.95	the paper's words
concepts	perturbation-based approaches Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: perturbation-based approaches	`concepts[126]`	0.95	the paper's words
concepts	back-propagation-based techniques Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: back-propagation-based techniques	`concepts[127]`	0.95	the paper's words
concepts	score of confidence Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: score of confidence	`concepts[128]`	0.95	the paper's words
concepts	temperature scaling Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: temperature scaling	`concepts[129]`	0.95	the paper's words
concepts	incorporating new information into a plain DL model Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: incorporating new information into a plain DL model	`concepts[130]`	0.95	the paper's words
concepts	parameter pruning Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: parameter pruning	`concepts[131]`	0.95	the paper's words
concepts	knowledge distillation Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: knowledge distillation	`concepts[132]`	0.95	the paper's words
concepts	compact convolution filters Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: compact convolution filters	`concepts[133]`	0.95	the paper's words
concepts	low-rank factorization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: low-rank factorization	`concepts[134]`	0.95	the paper's words
concepts	lack of training data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: lack of training data	`concepts[135]`	0.95	the paper's words
concepts	stress tests Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: stress tests	`concepts[136]`	0.95	the paper's words
concepts	classification Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: classification	`concepts[137]`	0.95	the paper's words
concepts	localization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: localization	`concepts[138]`	0.95	the paper's words
concepts	detection Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: detection	`concepts[139]`	0.95	the paper's words
concepts	segmentation Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: segmentation	`concepts[140]`	0.95	the paper's words
concepts	registration Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: registration	`concepts[141]`	0.95	the paper's words
concepts	Target Selection Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Target Selection	`concepts[142]`	0.95	the paper's words
concepts	Feature Extraction Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Feature Extraction	`concepts[143]`	0.95	the paper's words
concepts	Feature Matching Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Feature Matching	`concepts[144]`	0.95	the paper's words
concepts	Pose Optimization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Pose Optimization	`concepts[145]`	0.95	the paper's words
concepts	CPU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: CPU	`concepts[146]`	0.95	the paper's words
concepts	GPU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: GPU	`concepts[147]`	0.95	the paper's words
concepts	FPGA Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: FPGA	`concepts[148]`	0.95	the paper's words
concepts	Accuracy Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Accuracy	`concepts[149]`	0.95	the paper's words
concepts	Sensitivity or Recall Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sensitivity or Recall	`concepts[150]`	0.95	the paper's words
concepts	Specificity Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Specificity	`concepts[151]`	0.95	the paper's words
concepts	Precision Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Precision	`concepts[152]`	0.95	the paper's words
concepts	F1-Score Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: F1-Score	`concepts[153]`	0.95	the paper's words
concepts	J Score Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: J Score	`concepts[154]`	0.95	the paper's words
concepts	Youdens J statistic Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Youdens J statistic	`concepts[155]`	0.95	the paper's words
concepts	Area Under the ROC Curve Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Area Under the ROC Curve	`concepts[156]`	0.95	the paper's words
concepts	RNN category Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: RNN category	`concepts[157]`	0.95	the paper's words
concepts	increasing the stride value Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: increasing the stride value	`concepts[158]`	0.95	the paper's words
concepts	a feature map of lower dimensions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: a feature map of lower dimensions	`concepts[159]`	0.95	the paper's words
concepts	sub-sampling layers Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: sub-sampling layers	`concepts[160]`	0.95	the paper's words
concepts	a reduction in the network parameters Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: a reduction in the network parameters	`concepts[161]`	0.95	the paper's words
concepts	how confident the model is in its prediction Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: how confident the model is in its prediction	`concepts[162]`	0.95	the paper's words
concepts	overfitting problem Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: overfitting problem	`concepts[163]`	0.95	the paper's words
concepts	visual cortex Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: visual cortex	`concepts[164]`	0.95	the paper's words
concepts	exploding gradient and vanishing problems Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: exploding gradient and vanishing problems	`concepts[165]`	0.95	the paper's words
concepts	model compression Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: model compression	`concepts[166]`	0.95	the paper's words
concepts	DNN training Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: DNN training	`concepts[167]`	0.95	the paper's words
concepts	inference accelerators Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: inference accelerators	`concepts[168]`	0.95	the paper's words
concepts	Sensitivity Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sensitivity	`concepts[169]`	0.95	the paper's words
concepts	Recall Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Recall	`concepts[170]`	0.95	the paper's words
unhoused_relations	CNN → visual cortex Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. CNN structure is described as inspired by visual-cortex organization; inspired-by is absent from the eight governed types. — the relation stated: Convolutional neural networks Quotation withheld	`unhoused_relations[0]`	0.95	the tagger's words
unhoused_relations	LSTM → exploding gradient and vanishing problems Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. LSTM is presented as handling the recurrent network's exploding/vanishing-gradient issue; mitigates is absent from the eight governed types. — the relation stated: Recurrent neural networks Typical unfolded RNN diagram Quotation withheld	`unhoused_relations[1]`	0.95	the tagger's words
unhoused_relations	parameter pruning → model compression Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. Parameter pruning is one of four methods used for model compression; method-for is absent, and the prose does not support forcing an ontological part-of edge. — the relation stated: Model compression Quotation withheld	`unhoused_relations[2]`	0.95	the tagger's words
unhoused_relations	knowledge distillation → model compression Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. Knowledge distillation trains a compact model from a larger model for compression; method-for is absent from the governed types. — the relation stated: Model compression Quotation withheld	`unhoused_relations[3]`	0.95	the tagger's words
unhoused_relations	registration → Target Selection Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. Target Selection is the first named stage of image registration; stage-of is absent from the eight governed types. — the relation stated: Registration Quotation withheld	`unhoused_relations[4]`	0.95	the tagger's words
unhoused_relations	registration → Pose Optimization Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. Pose Optimization is the distance-minimization stage of image registration; stage-of is absent from the eight governed types. — the relation stated: Registration Quotation withheld	`unhoused_relations[5]`	0.95	the tagger's words
unhoused_relations	GPU → DNN training Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. The GPU is described as optimized for DNN training; optimized-for is absent from the eight governed types. — the relation stated: CPU-based approach Quotation withheld	`unhoused_relations[6]`	0.95	the tagger's words
unhoused_relations	FPGA → inference accelerators Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. FPGAs are commonly used to implement inference accelerators; used-for is absent from the governed relation types. — the relation stated: FPGA-based approach Quotation withheld	`unhoused_relations[7]`	0.95	the tagger's words
unhoused_relations	Sensitivity → Recall Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. Sensitivity and Recall are presented as alternate names for one metric; the slash/or naming construction lacks a governed same-as cue, so the alias is filed unhoused. — the relation stated: Evaluation metrics Quotation withheld	`unhoused_relations[8]`	0.95	the tagger's words
unhoused_relations	CapsuleNet → CapsNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. CapsNet is the short form used for CapsuleNet; abbreviation expansion is absent from the eight governed types. — the relation stated: CapsuleNet Quotation withheld	`unhoused_relations[9]`	0.95	the tagger's words
senses	RL - REPRESENTATION LEARNING. Given in the introduction as another name for deep learning itself Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[0]`	0.95	the tagger's words
senses	RL - REINFORCEMENT LEARNING. Given 16,670 characters later as another name for deep reinforcement learning, a distinct technique the same paper classifies separately. Neither definition acknowledges the other Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[1]`	0.95	the tagger's words
senses	transfer learning - AS RECOMMENDED: enhances accuracy across pattern recognition and computer vision, and is the paper's answer to insufficient training data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[2]`	0.95	the tagger's words
senses	transfer learning - AS MEASURED ACROSS DISSIMILAR DOMAINS: does not significantly affect performance, with models trained from scratch performing nearly as well. Same technique, opposite verdict, and the disambiguator is source-target similarity Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[3]`	0.95	the tagger's words
senses	Feature engineering versus deep representation learning - conventional ML constructs task-specific features manually, whereas DL extracts layered discriminative features automatically Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[4]`	0.95	the tagger's words
senses	Recurrent state - the hidden layer functions as short-term memory carrying information from earlier elements of a sequence Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[5]`	0.95	the tagger's words
senses	CNN locality - sparse local connections and shared weights exploit image structure while reducing parameter and memory requirements Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[6]`	0.95	the tagger's words
senses	Pooling trade-off - subsampling reduces parameters and computation but can discard the location of a detected feature Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[7]`	0.95	the tagger's words
senses	Activation-function family - ReLU reduces computation but can permanently deactivate neurons; leaky, noisy, and parametric variants modify that behavior Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[8]`	0.95	the tagger's words
senses	Fitting-state distinction - over-fitting fails on unseen data, under-fitting fails to learn enough, and just-fitting succeeds on both training and test sets Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[9]`	0.95	the tagger's words
senses	Dense connectivity - DenseNet concatenates all preceding feature maps, improving information flow at the cost of more feature maps and parameter expense Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[10]`	0.95	the tagger's words
senses	Uncertainty calibration - raw softmax probabilities may be overconfident and require post-scaling before they function as reliable confidence estimates Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[11]`	0.95	the tagger's words
senses	Compression taxonomy - parameter pruning, knowledge distillation, compact filters, and low-rank factorization reduce different sources of model cost Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[12]`	0.95	the tagger's words
senses	Application-task boundary - localization uses one bounding box, detection may use many, segmentation outlines pixels, and registration aligns images Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[13]`	0.95	the tagger's words
senses	Hardware specialization - CPU favors memory and generality, GPU favors parallel DNN training, and FPGA favors configurable low-latency inference Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[14]`	0.95	the tagger's words
senses	Metric-role distinction - evaluation metrics guide classifier selection during training and assess efficiency on hidden data during testing Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`senses[15]`	0.95	the tagger's words
claim_relations	critiques → existing DL review papers, each of which addresses a single application or topic and so leaves an overall lack of knowledge about the field Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[0]`	0.95	the tagger's words
claim_relations	builds-on → the 2020 Google identification of underspecification as a distinct failure mode of ML pipelines Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[1]`	0.95	the tagger's words
claim_relations	qualifies → transfer learning from ImageNet as the standard remedy for insufficient training data in medical imaging Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`claim_relations[2]`	0.95	the tagger's words
concept_edges	DL —part-of→ ML Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 0.95 · the relation stated: deep learning is a subset of machine learning also in: DL · ML	`concept_edges[0]`	0.95	the tagger's words
concept_edges	representation learning —same-as→ RL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 0.9 · the relation stated: given in apposition in the introduction also in: representation learning · RL	`concept_edges[1]`	0.95	the tagger's words
concept_edges	deep reinforcement learning —same-as→ RL Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 0.9 · the relation stated: given as another name 16,670 characters later, for a different concept also in: deep reinforcement learning · RL	`concept_edges[2]`	0.95	the tagger's words
concept_edges	TL —contrasted-with→ lightweight models trained from scratch Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. differs in: performance on medical imaging tasks when source and target domains differ · the relation itself is separately evidenced · strength of the relation: 0.9 · differs in (stated by the tagger, no quotation): performance on medical imaging tasks when source and target domains differ · the relation stated: cross-domain transfer does not beat training from scratch on medical imaging also in: TL · lightweight models trained from scratch	`concept_edges[3]`	0.95	the tagger's words
concept_edges	RNN —contrasted-with→ conventional networks Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The review explicitly contrasts recurrent use of sequential state with conventional networks. also in: RNN · conventional networks	`concept_edges[4]`	0.95	the tagger's words
concept_edges	CNN —contrasted-with→ RNN Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The paper directly compares their power and feature compatibility. also in: CNN · RNN	`concept_edges[5]`	0.95	the tagger's words
concept_edges	gated recurrent units (GRUs) —part-of→ RNN category Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The RNN category is explicitly said to include GRUs. also in: gated recurrent units (GRUs) · RNN category	`concept_edges[6]`	0.95	the tagger's words
concept_edges	long short-term memory (LSTM) approaches —part-of→ RNN category Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The RNN category is explicitly said to include LSTM approaches. also in: long short-term memory (LSTM) approaches · RNN category	`concept_edges[7]`	0.95	the tagger's words
concept_edges	convolutional filters —part-of→ convolutional layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The convolutional layer is stated to consist of a collection of filters. also in: convolutional filters · convolutional layer	`concept_edges[8]`	0.95	the tagger's words
concept_edges	max pooling —part-of→ pooling methods Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: Max pooling is explicitly included among pooling methods. also in: max pooling · pooling methods	`concept_edges[9]`	0.95	the tagger's words
concept_edges	global average pooling (GAP) —part-of→ pooling methods Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: GAP is explicitly included among pooling methods. also in: global average pooling (GAP) · pooling methods	`concept_edges[10]`	0.95	the tagger's words
concept_edges	increasing the stride value —causes→ a feature map of lower dimensions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The lower-dimensional map is stated to result from increasing stride. also in: increasing the stride value · a feature map of lower dimensions	`concept_edges[11]`	0.95	the tagger's words
concept_edges	sub-sampling layers —causes→ a reduction in the network parameters Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: Down-sampling is stated to lead to fewer parameters. also in: sub-sampling layers · a reduction in the network parameters	`concept_edges[12]`	0.95	the tagger's words
concept_edges	Cross-Entropy or Softmax Loss Function —same-as→ log loss function Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The review explicitly gives log loss as another name. also in: Cross-Entropy or Softmax Loss Function · log loss function	`concept_edges[13]`	0.95	the tagger's words
concept_edges	Drop-Weights —contrasted-with→ Dropout Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: Weights rather than neurons are dropped; the paper calls this the only difference. Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but Drop-Weights is 'highly similar to dropout', differing in what is dropped; contrasted-with is loose. also in: Drop-Weights · Dropout	`concept_edges[14]`	0.95	the tagger's words
concept_edges	Adam —is-a→ learning strategy Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The paper directly classifies Adam as a learning strategy. also in: Adam · learning strategy	`concept_edges[15]`	0.95	the tagger's words
concept_edges	ResNet —contrasted-with→ highway network Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The review contrasts ResNet's always-open parameter-free shortcuts with highway gates. also in: ResNet · highway network	`concept_edges[16]`	0.95	the tagger's words
concept_edges	High-Resolution Network (HRNet) —same-as→ HRNet Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The long name is explicitly followed by the short name. also in: High-Resolution Network (HRNet) · HRNet	`concept_edges[17]`	0.95	the tagger's words
concept_edges	catastrophic forgetting —operationalised-as→ incorporating new information into a plain DL model Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The paper gives an explicit definition. · the relation stated: Catastrophic forgetting Quotation withheld Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: 'This' = catastrophic forgetting (the heading immediately before); the definition is stated. also in: catastrophic forgetting · incorporating new information into a plain DL model	`concept_edges[18]`	0.95	the tagger's words
concept_edges	score of confidence —operationalised-as→ how confident the model is in its prediction Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The paper explicitly defines the confidence score. also in: score of confidence · how confident the model is in its prediction	`concept_edges[19]`	0.95	the tagger's words
concept_edges	FPGA —contrasted-with→ GPU Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The processors are directly compared on floating-point performance and configuration. also in: FPGA · GPU	`concept_edges[20]`	0.95	the tagger's words
concept_edges	J Score —same-as→ Youdens J statistic Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. the relation itself is separately evidenced · strength of the relation: 1 · note: The paper explicitly states the alternate metric name. also in: J Score · Youdens J statistic	`concept_edges[21]`	0.95	the tagger's words
content_tags	transfer learning from a dissimilar domain does not significantly affect medical imaging performance; models trained from scratch do nearly as well Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[1]`	0.95	the tagger's words
content_tags	underspecification means small modifications force a model toward a completely different solution and different deployment predictions Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[2]`	0.95	the tagger's words
content_tags	the remedy proposed for underspecification requires already understanding how the model can fail Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[3]`	0.95	the tagger's words
content_tags	CNN performance is greatly influenced by hyper-parameter selection; any small change affects it. Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[4]`	0.95	the tagger's words
content_tags	most current deep-learning models still use supervised learning despite the stated transition toward semi- and unsupervised methods Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.findings[5]`	0.95	the tagger's words
content_tags	a ranking-type metric that, in contrast to probability and threshold metrics, exposes the entire classifier ranking performance Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: AUC	`content_tags.definitions[0]`	0.95	the tagger's words
content_tags	a subset of ML, inspired by the information processing patterns of the human brain, which requires no human-designed rules and maps input to labels from data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: DL	`content_tags.definitions[1]`	0.95	the tagger's words
content_tags	only a few weights between two adjacent layers, against fully connected networks where every neuron links to every neuron in the next layer Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sparse Connectivity	`content_tags.definitions[2]`	0.95	the tagger's words
content_tags	no weights allocated between any two neurons of neighbouring layers; one set of weights operates over every pixel of the input matrix Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Weight Sharing	`content_tags.definitions[3]`	0.95	the tagger's words
content_tags	neurons randomly dropped during each training epoch, distributing feature-selection power across the whole group and forcing independent features Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Dropout	`content_tags.definitions[4]`	0.95	the tagger's words
content_tags	incorporating new information into a plain DL model by interfering with the learned information - fine-tuning on a new class breaks the old ones Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: catastrophic forgetting	`content_tags.definitions[5]`	0.85	the tagger's words
content_tags	the reason ML models show surprisingly poor behaviour when tested in real-world applications despite passing their own validation Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: underspecification	`content_tags.definitions[6]`	0.95	the tagger's words
content_tags	using images that look similar to the target dataset for training - e.g. X-rays of other chest diseases before fine-tuning on COVID-19 chest X-rays Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Same-domain TL	`content_tags.definitions[7]`	0.95	the tagger's words
content_tags	papers reviewed from 2020, of more than 300 total Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.quantities[0]`	0.95	the tagger's words
content_tags	the floor on the number of papers reviewed Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above.	`content_tags.quantities[1]`	0.95	the tagger's words
content_tags	AUC Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: AUC	`content_tags.statistics[0]`	0.95	the tagger's words
content_tags	Sensitivity or Recall Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Sensitivity or Recall	`content_tags.statistics[1]`	0.95	the tagger's words
content_tags	Specificity Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: Specificity	`content_tags.statistics[2]`	0.95	the tagger's words
content_tags	hyper-parameter selection Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: hyper-parameter selection	`content_tags.variables[0]`	0.95	the tagger's words
content_tags	amount of data Quotation withheld — this record is not cleared for redistribution. The sentence is in the paper at the link above. also in: amount of data	`content_tags.variables[1]`	0.95	the tagger's words

Interpretive choices

Tags that could not be placed in the text:

none — every tag was located in the text

What was deliberately not tagged, and why, is recorded per paper by the tagger. Where a paper has no such record, this page says so rather than implying that nothing was declined.