Similarity of Processing Steps in Vision Model Representations
Authors:
- Matéo MahautMMMatéo Mahaut
- Marco BaroniMBMarco Baroni
Preprint
Abstract
Recent literature suggests that the bigger the model, the more likely it is to converge to similar, ``universal'' representations, despite different training objectives, datasets, or modalities. While this literature shows that there is an area where model representations are similar, we study here how vision models might get to those representations -- in particular, do they also converge to the same intermediate steps and operations? We therefore study the processes that lead to convergent representations in different models. First, we quantify distance between different model representations at different stages. We follow the evolution of distances between models throughout processing, identifying the processing steps which are most different between models. We find that while layers at similar positions in different models have the most similar representations, strong differences remain. Classifier models, unlike the others, will discard information about low-level image statistics in their final layers. CNN- and transformer-based models also behave differently, with transformer models applying smoother changes to representations from one layer to the next. These distinctions clarify the level and nature of convergence between model representations, and enables a more qualitative account of the underlying processes in image models.Keywords
All Related Versions
This publication has 9 references indexed in Scilit:
- Bridging Information-Theoretic and Geometric Compression in Language Models2023
- Ranking the information content of distance measuresPNAS Nexus, 2022
- Visual ensemble representations in Deep Neural Networks trained for natural object recognitionJournal of Vision, 2021
- Convolutional Neural Networks as a Model of the Visual System: Past, Present, and FutureJournal of Cognitive Neuroscience, 2021
- Humans Meet Models on Object Naming: A New Dataset and Analysis2020
- ImageNet classification with deep convolutional neural networksCommunications of the ACM, 2017
- Using goal-driven deep learning models to understand sensory cortexNature Neuroscience, 2016
- ImageNet Large Scale Visual Recognition ChallengeInternational Journal of Computer Vision, 2015
- RELATIONS BETWEEN TWO SETS OF VARIATESBiometrika, 1936