Scale Alone Does not Improve Mechanistic Interpretability in Vision Models

• URL: http://arxiv.org/abs/2307.05471v2
• Date: Sat, 30 Mar 2024 16:06:36 GMT
• Title: Scale Alone Does not Improve Mechanistic Interpretability in Vision Models
• Authors: Roland S. Zimmermann, Thomas Klein, Wieland Brendel,
• Abstract summary: Machine vision has seen remarkable progress by scaling neural networks to unprecedented levels in dataset and model size. We quantify one form of mechanistic interpretability for a diverse suite of nine models. None of the investigated state-of-the-art models are easier to interpret than the GoogLeNet model from almost a decade ago.
• Score: 16.020535763297175
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In light of the recent widespread adoption of AI systems, understanding the internal information processing of neural networks has become increasingly critical. Most recently, machine vision has seen remarkable progress by scaling neural networks to unprecedented levels in dataset and model size. We here ask whether this extraordinary increase in scale also positively impacts the field of mechanistic interpretability. In other words, has our understanding of the inner workings of scaled neural networks improved as well? We use a psychophysical paradigm to quantify one form of mechanistic interpretability for a diverse suite of nine models and find no scaling effect for interpretability - neither for model nor dataset size. Specifically, none of the investigated state-of-the-art models are easier to interpret than the GoogLeNet model from almost a decade ago. Latest-generation vision models appear even less interpretable than older architectures, hinting at a regression rather than improvement, with modern models sacrificing interpretability for accuracy. These results highlight the need for models explicitly designed to be mechanistically interpretable and the need for more helpful interpretability methods to increase our understanding of networks at an atomic level. We release a dataset containing more than 130'000 human responses from our psychophysical evaluation of 767 units across nine models. This dataset facilitates research on automated instead of human-based interpretability evaluations, which can ultimately be leveraged to directly optimize the mechanistic interpretability of models.

Related papers

• Improving Neuron-level Interpretability with White-box Language Models [11.898535906016907]
  We introduce a white-box transformer-like architecture named Coding RAte TransformEr (CRATE) Our comprehensive experiments showcase significant improvements (up to 103% relative improvement) in neuron-level interpretability. CRATE's increased interpretability comes from its enhanced ability to consistently and distinctively activate on relevant tokens.
  arXiv  Detail & Related papers  (2024-10-21T19:12:33Z)
• Improving Network Interpretability via Explanation Consistency Evaluation [56.14036428778861]
  We propose a framework that acquires more explainable activation heatmaps and simultaneously increase the model performance. Specifically, our framework introduces a new metric, i.e., explanation consistency, to reweight the training samples adaptively in model learning. Our framework then promotes the model learning by paying closer attention to those training samples with a high difference in explanations.
  arXiv  Detail & Related papers  (2024-08-08T17:20:08Z)
• Data-efficient Large Vision Models through Sequential Autoregression [58.26179273091461]
  We develop an efficient, autoregression-based vision model on a limited dataset. We demonstrate how this model achieves proficiency in a spectrum of visual tasks spanning both high-level and low-level semantic understanding. Our empirical evaluations underscore the model's agility in adapting to various tasks, heralding a significant reduction in the parameter footprint.
  arXiv  Detail & Related papers  (2024-02-07T13:41:53Z)
• Visual Prompting Upgrades Neural Network Sparsification: A Data-Model Perspective [64.04617968947697]
  We introduce a novel data-model co-design perspective: to promote superior weight sparsity. Specifically, customized Visual Prompts are mounted to upgrade neural Network sparsification in our proposed VPNs framework.
  arXiv  Detail & Related papers  (2023-12-03T13:50:24Z)
• On Modifying a Neural Network's Perception [3.42658286826597]
  We propose a method which allows one to modify what an artificial neural network is perceiving regarding specific human-defined concepts. We test the proposed method on different models, assessing whether the performed manipulations are well interpreted by the models, and analyzing how they react to them.
  arXiv  Detail & Related papers  (2023-03-05T12:09:37Z)
• Neural Additive Models for Location Scale and Shape: A Framework for Interpretable Neural Regression Beyond the Mean [1.0923877073891446]
  Deep neural networks (DNNs) have proven to be highly effective in a variety of tasks. Despite this success, the inner workings of DNNs are often not transparent. This lack of interpretability has led to increased research on inherently interpretable neural networks.
  arXiv  Detail & Related papers  (2023-01-27T17:06:13Z)
• Dynamic Inference with Neural Interpreters [72.90231306252007]
  We present Neural Interpreters, an architecture that factorizes inference in a self-attention network as a system of modules. inputs to the model are routed through a sequence of functions in a way that is end-to-end learned. We show that Neural Interpreters perform on par with the vision transformer using fewer parameters, while being transferrable to a new task in a sample efficient manner.
  arXiv  Detail & Related papers  (2021-10-12T23:22:45Z)
• Sparse Flows: Pruning Continuous-depth Models [107.98191032466544]
  We show that pruning improves generalization for neural ODEs in generative modeling. We also show that pruning finds minimal and efficient neural ODE representations with up to 98% less parameters compared to the original network, without loss of accuracy.
  arXiv  Detail & Related papers  (2021-06-24T01:40:17Z)
• Are Convolutional Neural Networks or Transformers more like human vision? [9.83454308668432]
  We show that attention-based networks can achieve higher accuracy than CNNs on vision tasks. These results have implications both for building more human-like vision models, as well as for understanding visual object recognition in humans.
  arXiv  Detail & Related papers  (2021-05-15T10:33:35Z)
• Firearm Detection via Convolutional Neural Networks: Comparing a Semantic Segmentation Model Against End-to-End Solutions [68.8204255655161]
  Threat detection of weapons and aggressive behavior from live video can be used for rapid detection and prevention of potentially deadly incidents. One way for achieving this is through the use of artificial intelligence and, in particular, machine learning for image analysis. We compare a traditional monolithic end-to-end deep learning model and a previously proposed model based on an ensemble of simpler neural networks detecting fire-weapons via semantic segmentation.
  arXiv  Detail & Related papers  (2020-12-17T15:19:29Z)
• The Self-Simplifying Machine: Exploiting the Structure of Piecewise Linear Neural Networks to Create Interpretable Models [0.0]
  We introduce novel methodology toward simplification and increased interpretability of Piecewise Linear Neural Networks for classification tasks. Our methods include the use of a trained, deep network to produce a well-performing, single-hidden-layer network without further training. On these methods, we conduct preliminary studies of model performance, as well as a case study on Wells Fargo's Home Lending dataset.
  arXiv  Detail & Related papers  (2020-12-02T16:02:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.