Representational Similarity

Representational similarity analysis (RSA) investigates how similar the internal representations of information are across different systems, such as brains and artificial neural networks. Current research focuses on developing and applying RSA methods to compare representations across various model architectures (e.g., convolutional neural networks, transformers, generative models), assessing the robustness and generalizability of these representations, and exploring their relationship to cognitive processes and task performance. This work is significant for advancing our understanding of both biological and artificial intelligence, potentially leading to more robust and human-aligned AI systems and improved neuroscientific models of brain function.

Papers

July 19, 2022

Similarity of Pre-trained and Fine-tuned Representations
Thomas Goerttler, Klaus Obermayer
Domain Adaptation Transfer Learning Pre Trained Shot Learning High Similarity Representational Similarity Representation Finetuning

July 15, 2022

Probing Semantic Grounding in Language Models of Code with Representational Similarity Analysis
Shounak Naik, Rajaswa Patil, Swati Agarwal, Veeky Baths
Language Model Real World Code Representational Similarity Input Modality Semantic Grounding Semantic Perturbation

June 23, 2022

Measuring Representational Robustness of Neural Networks Through Shared Invariances
Vedant Nanda, Till Speicher, Camila Kolling, John P. Dickerson, Krishna P. Gummadi, Adrian Weller
Neural Network Deep Learning Native Robustness Alpha Invariance Robust Representation Representational Similarity Value Set Invariance

June 22, 2022

Coupling Visual Semantics of Artificial Neural Networks and Human Brain Function via Synchronized Activations
Lin Zhao, Haixing Dai, Zihao Wu, Zhenxiang Xiao, Lu Zhang, David Weizhong Liu, Xintao Hu, Xi Jiang, Sheng Li, Dajiang Zhu, Tianming Liu
Neural Network Visual Representation Brain Network Targeted Activation Penalty Brain Inspired Representational Similarity Visual Semantics

May 17, 2022

How do Variational Autoencoders Learn? Insights from Representational Similarity
Lisa Bonheme, Marek Grzes
Variational Autoencoder Variational Autoencoders Meaningful Representation DCU Insight AQ Encoder Side Representational Similarity Variance Network

January 31, 2022

Deconfounded Representation Similarity for Comparison of Neural Networks
Tianyu Cui, Yogesh Kumar, Pekka Marttinen, Samuel Kaski
Neural Network Transfer Learning Consistent Comparison Representational Similarity Similar Representation Domain Similarity

December 3, 2021

Divergent representations of ethological visual inputs emerge from supervised, unsupervised, and reinforcement learning
Grace W. Lindsay, Josh Merel, Tom Mrsic-Flogel, Maneesh Sahani
Reinforcement Learning Neural Representation Internal Representation Visual Input Representational Similarity Reinforcement Learning Task Dissimilar Representation

November 3, 2021

Drop, Swap, and Generate: A Self-Supervised Approach for Generating Neural Activity
Ran Liu, Mehdi Azabou, Max Dabagia, Chi-Heng Lin, Mohammad Gheshlaghi Azar, Keith B. Hengen, Michal Valko, Eva L. Dyer
Neural Activity Neural Recording Generate Quick Representational Similarity Self Supervised Approach Swap Distance Minimization Neural Generation Neural State

Representational Similarity

Papers

Similarity of Pre-trained and Fine-tuned Representations

Probing Semantic Grounding in Language Models of Code with Representational Similarity Analysis

Measuring Representational Robustness of Neural Networks Through Shared Invariances

Coupling Visual Semantics of Artificial Neural Networks and Human Brain Function via Synchronized Activations

How do Variational Autoencoders Learn? Insights from Representational Similarity

Deconfounded Representation Similarity for Comparison of Neural Networks

Divergent representations of ethological visual inputs emerge from supervised, unsupervised, and reinforcement learning

Drop, Swap, and Generate: A Self-Supervised Approach for Generating Neural Activity