{"id":"arxiv:2007.02789","title":"Comparing representational geometries using whitened unbiased-distance-matrix similarity","date":"2020-07-06","section":"Representation alignment","quality":0.698414,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/","source_url":"https://arxiv.org/abs/2007.02789","text_available":true,"authors":["Jörn Diedrichsen","Eva Berlot","Marieke Mur","Heiko H. Schütt"],"outgoing":[],"incoming":[{"source":"arxiv:2602.14486","target":"arxiv:2007.02789","evidence":"title","id":"arxiv:2602.14486","title":"Revisiting the Platonic Representation Hypothesis: An Aristotelian View","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/7f56b2d833094d17524bf9ad90c5d16bc602e9888589afd624ae8efc3f9d9fc0/"}],"text_parts":[{"start":0,"end":15310,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html"},{"start":15310,"end":30887,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html"},{"start":30887,"end":46745,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html"},{"start":46745,"end":62698,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html"},{"start":62698,"end":78232,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html"},{"start":78232,"end":85328,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-6.html"}],"contents":[{"title":"Comparing representational geometries using whitened unbiased-distance-matrix similarity","offset":368,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"Abstract","offset":460,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"keywords","offset":2169,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"1 Introduction","offset":2186,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"2 Results","offset":8252,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"2.1 Basic Definitions","offset":8266,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"2.2 Biased and unbiased estimates for the squared Euclidean distance","offset":11351,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"2.3 Variance of distance estimates","offset":14015,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-1.html#text"},{"title":"2.4 Model comparison using RDM correlations or cosine similarity","offset":17235,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.5 Using RDM cosine similarity prevents bias in model selection","offset":20072,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.6 RDM cosine similarity exploits informative zero-point","offset":23401,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.7 Covariance of distance estimates","offset":25351,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.8 RDM comparison in whitened RDM-space","offset":28044,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.9 Influence of RDM whitening on model comparisons","offset":30830,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-2.html#text"},{"title":"2.10 Factors influencing the advantage of RDM whitening","offset":33275,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"2.11 Interpretation of the WUC","offset":35770,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"2.12 Extension of the WUC to flexible RDM models","offset":36933,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"2.13 Relationship to other multivariate dependence measures","offset":38043,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"3 Discussion","offset":43068,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"3.1 Dealing with bias and covariance of distance estimators","offset":43732,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"3.2 When should the whitened unbiased RDM cosine similarity be used?","offset":46671,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-3.html#text"},{"title":"3.3 Probabilistic versions of RSA","offset":49374,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"3.4 Influence of spatial noise correlations","offset":50549,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"3.5 Electrophysiological and other non-Gaussian data","offset":52845,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"3.6 Summary","offset":54353,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4 Methods","offset":55044,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4.1 Extended definitions","offset":55058,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4.2 Bias of distance estimates","offset":57895,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4.3 Variance of distance estimates","offset":59017,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4.4 Spatial pre-whitening and Mahalanobis distances","offset":61455,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-4.html#text"},{"title":"4.5 Simulations","offset":65426,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"5 acknowledgements","offset":69669,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6 Appendix","offset":70034,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6.1 Derivation of variance-covariance matrix of the distance estimate","offset":70049,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6.1.1 Expectations of products of random variables","offset":70124,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6.1.2 Expectations of the product of normal matrices","offset":71163,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6.1.3 Averaging across partitions","offset":73723,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-5.html#text"},{"title":"6.2 Unbalanced designs","offset":79570,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-6.html#text"},{"title":"6.3 Simulations with spatially correlated noise","offset":82238,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-6.html#text"},{"title":"Instructions for reporting errors","offset":84364,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/part-6.html#text"}],"text_code_points":85328,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1c8778896169013495269ae2f98737d8338cf63d6943e50f134deef61ddf7e62/source.md"}