Making Sense Of Distributed Representations With Activation Spectroscopy

• URL: http://arxiv.org/abs/2501.15435v1
• Date: Sun, 26 Jan 2025 07:33:42 GMT
• Title: Making Sense Of Distributed Representations With Activation Spectroscopy
• Authors: Kyle Reing, Greg Ver Steeg, Aram Galstyan,
• Abstract summary: There is growing evidence to suggest that relevant features are encoded across many neurons in a distributed fashion. This work explores one feasible path to both detecting and tracing the joint influence of neurons in a distributed representation.
• Score: 44.94093096989921
• License:
• Abstract: In the study of neural network interpretability, there is growing evidence to suggest that relevant features are encoded across many neurons in a distributed fashion. Making sense of these distributed representations without knowledge of the network's encoding strategy is a combinatorial task that is not guaranteed to be tractable. This work explores one feasible path to both detecting and tracing the joint influence of neurons in a distributed representation. We term this approach Activation Spectroscopy (ActSpec), owing to its analysis of the pseudo-Boolean Fourier spectrum defined over the activation patterns of a network layer. The sub-network defined between a given layer and an output logit is cast as a special class of pseudo-Boolean function. The contributions of each subset of neurons in the specified layer can be quantified through the function's Fourier coefficients. We propose a combinatorial optimization procedure to search for Fourier coefficients that are simultaneously high-valued, and non-redundant. This procedure can be viewed as an extension of the Goldreich-Levin algorithm which incorporates additional problem-specific constraints. The resulting coefficients specify a collection of subsets, which are used to test the degree to which a representation is distributed. We verify our approach in a number of synthetic settings and compare against existing interpretability benchmarks. We conclude with a number of experimental evaluations on an MNIST classifier, and a transformer-based network for sentiment analysis.

Related papers

• Learning local discrete features in explainable-by-design convolutional neural networks [0.0]
  We introduce an explainable-by-design convolutional neural network (CNN) based on the lateral inhibition mechanism. The model consists of the predictor, that is a high-accuracy CNN with residual or dense skip connections. By collecting observations and directly calculating probabilities, we can explain causal relationships between motifs of adjacent levels.
  arXiv  Detail & Related papers  (2024-10-31T18:39:41Z)
• Spectral complexity of deep neural networks [2.099922236065961]
  We use the angular power spectrum of the limiting field to characterize the complexity of the network architecture. On this basis, we classify neural networks as low-disorder, sparse, or high-disorder. We show how this classification highlights a number of distinct features for standard activation functions, and in particular, sparsity properties of ReLU networks.
  arXiv  Detail & Related papers  (2024-05-15T17:55:05Z)
• Going Beyond Neural Network Feature Similarity: The Network Feature Complexity and Its Interpretation Using Category Theory [64.06519549649495]
  We provide the definition of what we call functionally equivalent features. These features produce equivalent output under certain transformations. We propose an efficient algorithm named Iterative Feature Merging.
  arXiv  Detail & Related papers  (2023-10-10T16:27:12Z)
• Permutation Equivariant Neural Functionals [92.0667671999604]
  This work studies the design of neural networks that can process the weights or gradients of other neural networks. We focus on the permutation symmetries that arise in the weights of deep feedforward networks because hidden layer neurons have no inherent order. In our experiments, we find that permutation equivariant neural functionals are effective on a diverse set of tasks.
  arXiv  Detail & Related papers  (2023-02-27T18:52:38Z)
• Probabilistic Verification of ReLU Neural Networks via Characteristic Functions [11.489187712465325]
  We use ideas from probability theory in the frequency domain to provide probabilistic verification guarantees for ReLU neural networks. We interpret a (deep) feedforward neural network as a discrete dynamical system over a finite horizon. We obtain the corresponding cumulative distribution function of the output set, which can be used to check if the network is performing as expected.
  arXiv  Detail & Related papers  (2022-12-03T05:53:57Z)
• Decomposing neural networks as mappings of correlation functions [57.52754806616669]
  We study the mapping between probability distributions implemented by a deep feed-forward network. We identify essential statistics in the data, as well as different information representations that can be used by neural networks.
  arXiv  Detail & Related papers  (2022-02-10T09:30:31Z)
• A research framework for writing differentiable PDE discretizations in JAX [3.4389358108344257]
  Differentiable simulators are an emerging concept with applications in several fields, from reinforcement learning to optimal control. We propose a library of differentiable operators and discretizations, by representing operators as mappings between families of continuous functions, parametrized by finite vectors. We demonstrate the approach on an acoustic optimization problem, where the Helmholtz equation is discretized using Fourier spectral methods, and differentiability is demonstrated using gradient descent to optimize the speed of sound of an acoustic lens.
  arXiv  Detail & Related papers  (2021-11-09T15:58:44Z)
• Linear approximability of two-layer neural networks: A comprehensive analysis based on spectral decay [4.042159113348107]
  We first consider the case of single neuron and show that the linear approximability, quantified by the Kolmogorov width, is controlled by the eigenvalue decay of an associate kernel. We show that similar results also hold for two-layer neural networks.
  arXiv  Detail & Related papers  (2021-08-10T23:30:29Z)
• Adversarial Examples Detection with Bayesian Neural Network [57.185482121807716]
  We propose a new framework to detect adversarial examples motivated by the observations that random components can improve the smoothness of predictors. We propose a novel Bayesian adversarial example detector, short for BATer, to improve the performance of adversarial example detection.
  arXiv  Detail & Related papers  (2021-05-18T15:51:24Z)
• Spatially Adaptive Inference with Stochastic Feature Sampling and Interpolation [72.40827239394565]
  We propose to compute features only at sparsely sampled locations. We then densely reconstruct the feature map with an efficient procedure. The presented network is experimentally shown to save substantial computation while maintaining accuracy over a variety of computer vision tasks.
  arXiv  Detail & Related papers  (2020-03-19T15:36:31Z)

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.