SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1905.00414 · 2019

Similarity of Neural Network Representations Revisited

Honglak Lee, Simon Kornblith, Mohammad Norouzi, Geoffrey Hinton

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 28 functions out of this paper's own repositories and ran 25 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

FunctionStatusWhere it lives
CKA Ran jayroxis/CKA-similarity/CKA.py
pointer only (licence: NONE) · get_code("d871d94be50b9e08")
CKA Ran lemonadec/Relevance-between-Accuracy-under-Black-box-Attack-and-the-Similarity-between-Networks/similarity.py
pointer only (licence: NONE) · get_code("ee46af7585e6a68f")
ModelInfo Ran ristoale97/centered-kernel-alignment/ckatorch/cka.py
code served (permissive licence) · get_code("1926e59186ecc073")
center_gram Ran this paper's copy was not recorded; identical code first harvested from Alxmrphi/correcting_CKA_alignment
pointer only · get_code("7a3daac7c3a534a6")
centering Ran yl2488/CKA-Centered-Kernel-Alignment/CKA.py
code served (permissive licence) · get_code("5c273c526536356d")
cka Ran this paper's copy was not recorded; identical code first harvested from Alxmrphi/correcting_CKA_alignment
pointer only · get_code("e9f80de46ff7dcd7")
cka_batch Ran ristoale97/centered-kernel-alignment/ckatorch/cka.py
code served (permissive licence) · get_code("7dc891b50b8982a5")
compute_CKA_similarity Ran kexueyi/discover_infant_vis/src/utils/cka_utils.py
pointer only (licence: NONE) · get_code("377bcf14c19d449e")
compute_ccas Ran yl2488/CKA-Centered-Kernel-Alignment/cca_core.py
code served (permissive licence) · get_code("01472465bed8b268")
compute_ccas Ran uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("9e9090cf436b48e6")
compute_corr Ran IBM/NeuronAlignment/utils/crosscorrelation.py
code served (permissive licence) · get_code("9737d6197d04f626")
create_zero_dict Ran uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("290ba45d53d47533")
get_activations Ran IBM/NeuronAlignment/utils/crosscorrelation.py
code served (permissive licence) · get_code("ca7973c5cf3fa73e")
get_zscore Ran IBM/NeuronAlignment/utils/crosscorrelation.py
code served (permissive licence) · get_code("d49541af704a1ba1")
hsic1 Ran ristoale97/centered-kernel-alignment/ckatorch/cka.py
code served (permissive licence) · get_code("9b1aad174038093c")
kernel_CKA Ran yuanli2333/CKA-Centered-Kernel-Alignment/CKA.py
code served (permissive licence) · get_code("395716aee7f73b45")
kernel_HSIC Ran yl2488/CKA-Centered-Kernel-Alignment/CKA.py
code served (permissive licence) · get_code("c633017bfba675e4")
linear_CKA Ran benjamin-mlr/first-align-then-predict/hidden_similarity/cka.py
pointer only (licence: NONE) · get_code("98db9cd60f9666cc")
linear_HSIC Ran benjamin-mlr/first-align-then-predict/hidden_similarity/cka.py
pointer only (licence: NONE) · get_code("9bcdcb150b92d09a")
positivedef_matrix_sqrt Ran yl2488/CKA-Centered-Kernel-Alignment/cca_core.py
code served (permissive licence) · get_code("ea98389a1755bb3f")
positivedef_matrix_sqrt Ran uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("3bd8381fd601f386")
rbf Ran yl2488/CKA-Centered-Kernel-Alignment/CKA.py
code served (permissive licence) · get_code("0f137cab706f1953")
remove_small Ran yl2488/CKA-Centered-Kernel-Alignment/cca_core.py
code served (permissive licence) · get_code("ffcbd13c7094948e")
remove_small Ran uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("f6ad5c8be57f992d")
sum_threshold Ran uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("1d8a61b6c896525a")
CKA Not yet run ristoale97/centered-kernel-alignment/ckatorch/cka.py
code served (permissive licence) · get_code("a86b6c3e6909ec10")
get_cca_similarity Not yet run uchicago-sandlab/robust_representation_similarity/similarity_metrics/cca_core.py
pointer only (licence: NONE) · get_code("fbc3e2408fc37570")
plot_cka Not yet run ristoale97/centered-kernel-alignment/ckatorch/cka.py
code served (permissive licence) · get_code("3f6aa279cc980040")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Recent work has sought to understand the behavior of neural networks by comparing representations between layers and between different trained models. We examine methods for comparing neural network representations based on canonical correlation analysis (CCA). We show that CCA belongs to a family of statistics for measuring multivariate similarity, but that neither CCA nor any other statistic that is invariant to invertible linear transformation can measure meaningful similarities between representations of higher dimension than the number of data points. We introduce a similarity index that measures the relationship between representational similarity matrices and does not suffer from this limitation. This similarity index is equivalent to centered kernel alignment (CKA) and is also closely connected to CCA. Unlike CCA, CKA can reliably identify correspondences between representations in networks trained from different initializations.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("1905.00414")
get_code_for_paper("1905.00414")
have("1905.00414")

Connect an agent — have() is free.