phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | spike sorting of high-channel-count extracellular recordings— “When multiple neurons close to the electrode fire action potentials, their spikes must be identified and assigned to the correct neuron based on the features of the recorded waveforms, a process known as spike sorting 1–15.” | |
| claim-type | method— “Here we introduce Kilosort, a spike sorting framework that meets these criteria, and show that it allows rapid and accurate sorting of large-scale in vivo data.” | |
| method — evidence | — “Here we overcome these limitations and present Kilosort, a new algorithm which takes advantage of a novel mathematical approach that greatly reduces the amount of calculation required, together with the computing capabilities of low-cost commercially available graphics processing units (GPUs).” | |
| mechanism | Replacing the full matrix convolution by a rank-3 low-rank template decomposition cuts the number of filtering operations by the ratio of channel count to template rank, which is what makes near-realtime sorting of hundreds of channels possible— “However, the low-rank decomposition of templates allows us to reduce the number of operations by a factor of Nchan/Nrank, where Nchan is the number of channels (typically > 100) and Nrank is the rank of the decomposed template (typically 3).” | mechanisms[0] |
| mechanism | Over-split clusters arise because the fitting algorithm splits a high-variance cluster across several templates in order to reduce total reconstruction error, which is why the resulting cluster pairs form one continuous density and can be merged automatically— “Instead, these clusters arise as a consequence of the algorithm partitioning clusters with large variance into multiple templates, so as to better explain their total variance.” | mechanisms[1] |
| extends | a previous template-matching method (numbered reference 6)— “This approach is related to a previous method 6, but that method requires a generic convex optimization that is slow for recordings with large numbers of channels.” | claim_relations[0] |
| extends | masked KlustaKwik (numbered reference 18)— “As we demonstrate with constructed ground-truth datasets, our system is more accurate than a current widely-used method 18.” | claim_relations[1] |
| builds-on | the hybrid ground truth benchmarking technique (numbered reference 18)— “We compared Kilosort and Klustakwik on 32 and 128 channel recordings, using a technique known as “hybrid ground truth” 18 (Fig. 3a-c).” | claim_relations[2] |
| sense | ground truth, here - MANUFACTURED, NOT OBSERVED: the paper's own term is 'hybrid ground truth', and the data set is made by taking human-curated clusters from a previous analysis, denoising their waveforms and re-inserting them elsewhere in a real recording, so the target the sorters must recover is a cleaned copy of an earlier judgement— “To create a hybrid ground truth data set, we first selected all the clusters from a recording that had been previously analyzed with KlustaKwik, and curated by a human expert” | senses[0] |
| sense | the reference standard's provenance - THE COMPETITOR'S OUTPUT: the clusters come from a recording previously analysed with KlustaKwik, which is the algorithm Kilosort is being compared against; re-insertion at new locations makes detection a fresh problem, but the standard is still derived from one of the things measured against it— “To create a hybrid ground truth data set, we first selected all the clusters from a recording that had been previously analyzed with KlustaKwik, and curated by a human expert” | senses[1] |
| sense | the reported score - GIVEN TWICE, ONCE AFTER AN ORACLE MERGE: the same panels are repeated after greedy best possible merges, with the number of merges needed to reach that best score shown separately, so one of the two numbers is an upper bound rather than an outcome— “Same as (def) after greedy best possible merges. j, Number of merges required to reach best score” | senses[2] |
| control | the denoised waveform SUBTRACTED from its original location after being added elsewhere — held — rules out an artificial increase in spike density at any point on the probe - a hybrid-data test that adds spikes without removing them changes the very statistic the sorter is being judged on— “To avoid increasing the spike density at any location on the probe, we also subtracted off the denoised waveform from its original location.” | controls[0] |
| outcome (stated by the tagger, no quotation) | control — held | controls[0] |
| for which use | limitation — any claim that the pipeline is automatic: OPERATOR CURATION IS STILL REQUIRED, PRIMARILY DUE TO NON-STATIONARITIES IN THE RECORDINGS SUCH AS ELECTRODE DRIFT— “Although we have defined a heuristic which eliminates the need for many manual merges, operator curation is still required, primarily due to non-stationarities in the recordings such as electrode drift.” | limitations[0] |
| for which use | limitation — the reproducibility of a given sort: LIKE MOST CLUSTERING ALGORITHMS, THIS MODEL IS PRONE TO NON-OPTIMAL LOCAL MINIMA— “Like most clustering algorithms, this model is prone to non-optimal local minima.” | limitations[1] |
| for which use | limitation — the speed comparison as a guide at the largest channel counts: the established framework NEEDED 480 MINUTES TO SORT THE 32 CHANNEL RECORDING, AND 15,000 MINUTES (10 DAYS) TO SORT THE 128 CHANNEL RECORDING, and no 384-channel comparison was run— “These were significant improvements over an established framework (masked KlustaKwik, 18), which needed 480 minutes to sort the 32 channel recording, and 15,000 minutes (10 days) to sort the 128 channel recording, while running on a CPU cluster (we did not attempt to run KlustaKwik on 384 channel recordings).” | limitations[2] |
| for which use | limitation — the accuracy comparison reported here: the hybrid ground truth data set was built from all the clusters of a recording previously analysed with KlustaKwik and curated by a human expert, and KlustaKwik is the algorithm Kilosort is compared against— “To create a hybrid ground truth data set, we first selected all the clusters from a recording that had been previously analyzed with KlustaKwik, and curated by a human expert” | limitations[3] |
| for which use | limitation — the 32-channel accuracy results: the authors detail only the 32 channel benchmarks, which they created by cropping the full 128 channel hybrid benchmark— “We detail here only the 32 channel benchmarks, which we created by cropping the full 128 channel hybrid benchmark (Fig. 3a-c), which we were able to run with many ground truth neurons” | limitations[4] |
| for which use | limitation — any score quoted from this figure: panels g-i repeat the results after greedy best possible merges, and a further panel gives the number of merges required to reach the best score— “Same as (def) after greedy best possible merges. j, Number of merges required to reach best score” | limitations[5] |
| replication | untested | |
| limitation | Manual operator curation is still needed despite the automatic merging heuristic, mainly because of electrode drift and other non-stationarities.— “Although we have defined a heuristic which eliminates the need for many manual merges, operator curation is still required, primarily due to non-stationarities in the recordings such as electrode drift.” | limitations[0] |
| limitation | The optimisation can settle in non-optimal local minima, which the authors only ameliorate by annealing several parameters. | limitations[1] |
| limitation | The speed comparison against the established framework was not run at the largest channel count, so no 384-channel baseline exists. | limitations[2] |
| limitation | The ground-truth spikes are the curated output of one of the two algorithms being compared. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[3] |
| limitation | The accuracy benchmark was made by cropping a larger one. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[4] |
| limitation | Scores are also reported after a best-possible merging step that no user could perform. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | limitations[5] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | Kilosort represents the raw voltage as a sum of template waveforms triggered at spike times, which allows spatiotemporally overlapping spikes to be identified and resolved rather than missed. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[0] |
| finding | The established KlustaKwik framework needed 480 minutes to sort a 32-channel recording and 15,000 minutes a 128-channel one on a CPU cluster, and Kilosort was a significant improvement on that. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[1] |
| finding | On hybrid ground-truth data Kilosort yielded well-isolated units (score above 0.9) for 53% of ground-truth neurons versus 35% for KlustaKwik. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[2] |
| finding | After greedy best-possible operator merges Kilosort still led, 69% of units above score 0.9 against 60% for KlustaKwik. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[3] |
| finding | Kilosort's accuracy advantage came from missing fewer spikes rather than from fewer false positives, which the authors attribute to its resolution of overlapping spikes. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[4] |
| finding | Fitting each spike with its own three-component spatiotemporal SVD ("private PCs") left about five times less residual waveform variance than one common per-channel PCA basis. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[5] |
| finding | Kilosort's run time scales linearly with the number of recorded neurons rather than the number of channels. Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.findings[6] |
| quantity | 10, 29 and 140 minutes — Kilosort run times for 32-, 128- and 384-channel recordings on a single GTX 980Ti workstation Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[0] |
| quantity | 480 minutes / 15,000 minutes (10 days) — KlustaKwik run times for the 32-channel and 128-channel recordings, the comparison baseline Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[1] |
| quantity | 53% vs 35% — share of ground-truth neurons sorted with a score above 0.9 by Kilosort versus KlustaKwik before manual merges Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[2] |
| quantity | 69% vs 60% — share of units scoring above 0.9 for Kilosort versus KlustaKwik after best possible merges Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[3] |
| quantity | ∼5 times — reduction in residual waveform variance from per-spike private PCs relative to a single per-channel PCA Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[4] |
| quantity | 16 — number of hybrid ground-truth dataset instantiations on which both algorithms were run Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[5] |
| quantity | Nchan/Nrank — factor by which the low-rank template decomposition reduces filtering operations, with channel count typically above 100 and template rank typically 3 Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.quantities[6] |
| variable | run time (dependent) Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[2] |
| variable | spike density (control) Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[4] |
| variable | miss rate (dependent) Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[5] |
| variable | false positive rate (dependent) Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.variables[6] |
| definition | Kilosort Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[0] |
| definition | spike sorting Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[1] |
| definition | common average referencing Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[2] |
| definition | zero-phase component analysis (ZCA) Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[3] |
| definition | hybrid ground truth Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[4] |
| definition | private PCs Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.definitions[5] |
| concept | hybrid ground truth Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[0] |
| concept | KlustaKwik Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[1] |
| concept | human expert Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[2] |
| concept | greedy best possible merges Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[3] |
| concept | ground truth neurons Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | concepts[4] |
| singular value decomposition | as written: SVD decomposition Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[0] |
| k-means clustering | as written: scaled K-means algorithm Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[1] |
| covariance matrix estimation Σ | as written: estimate the covariance matrix Σ Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[2] |
| Gaussian distribution | as written: Gaussian distribution whose variance scales with the square of its mean Quotation withheld — this paper is not ours to reproduce, and the page has already quoted as much of it as the rights layer allows. The tag, its field and its confidence stand; the sentence is at the publisher. | content_tags.statistics[3] |
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.
Tag layer © the Meta-Tagging Project, CC BY 4.0. The paper text is © its authors/publisher
and is linked, not reproduced. Excerpts shown are short quotations for identification and verification.
· back to index
· concepts · authors
· עברית
© 2026 Shir Sivroni · ORCID 0009-0000-2597-2824