Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decomposition

• URL: http://arxiv.org/abs/2501.14926v4
• Date: Fri, 07 Feb 2025 19:22:32 GMT
• Title: Interpretability in Parameter Space: Minimizing Mechanistic Description Length with Attribution-based Parameter Decomposition
• Authors: Dan Braun, Lucius Bushnaq, Stefan Heimersheim, Jake Mendel, Lee Sharkey,
• Abstract summary: We introduce a conceptual foundation for Attribution-based Decomposition (APD) APD directly decomposes a neural network's parameters into components that are faithful to the parameters of the original network. We demonstrate APD's effectiveness by successfully identifying ground truth mechanisms in toy experimental settings.
• Score: 0.0
• License:
• Abstract: Mechanistic interpretability aims to understand the internal mechanisms learned by neural networks. Despite recent progress toward this goal, it remains unclear how best to decompose neural network parameters into mechanistic components. We introduce Attribution-based Parameter Decomposition (APD), a method that directly decomposes a neural network's parameters into components that (i) are faithful to the parameters of the original network, (ii) require a minimal number of components to process any input, and (iii) are maximally simple. Our approach thus optimizes for a minimal length description of the network's mechanisms. We demonstrate APD's effectiveness by successfully identifying ground truth mechanisms in multiple toy experimental settings: Recovering features from superposition; separating compressed computations; and identifying cross-layer distributed representations. While challenges remain to scaling APD to non-toy models, our results suggest solutions to several open problems in mechanistic interpretability, including identifying minimal circuits in superposition, offering a conceptual foundation for 'features', and providing an architecture-agnostic framework for neural network decomposition.

Related papers

• Task-Oriented Real-time Visual Inference for IoVT Systems: A Co-design Framework of Neural Networks and Edge Deployment [61.20689382879937]
  Task-oriented edge computing addresses this by shifting data analysis to the edge. Existing methods struggle to balance high model performance with low resource consumption. We propose a novel co-design framework to optimize neural network architecture.
  arXiv  Detail & Related papers  (2024-10-29T19:02:54Z)
• Semantic Loss Functions for Neuro-Symbolic Structured Prediction [74.18322585177832]
  We discuss the semantic loss, which injects knowledge about such structure, defined symbolically, into training. It is agnostic to the arrangement of the symbols, and depends only on the semantics expressed thereby. It can be combined with both discriminative and generative neural models.
  arXiv  Detail & Related papers  (2024-05-12T22:18:25Z)
• Adaptive Multilevel Neural Networks for Parametric PDEs with Error Estimation [0.0]
  A neural network architecture is presented to solve high-dimensional parameter-dependent partial differential equations (pPDEs) It is constructed to map parameters of the model data to corresponding finite element solutions. It outputs a coarse grid solution and a series of corrections as produced in an adaptive finite element method (AFEM)
  arXiv  Detail & Related papers  (2024-03-19T11:34:40Z)
• Learning k-Level Structured Sparse Neural Networks Using Group Envelope Regularization [4.0554893636822]
  We introduce a novel approach to deploy large-scale Deep Neural Networks on constrained resources. The method speeds up inference time and aims to reduce memory demand and power consumption.
  arXiv  Detail & Related papers  (2022-12-25T15:40:05Z)
• Learning to Learn with Generative Models of Neural Network Checkpoints [71.06722933442956]
  We construct a dataset of neural network checkpoints and train a generative model on the parameters. We find that our approach successfully generates parameters for a wide range of loss prompts. We apply our method to different neural network architectures and tasks in supervised and reinforcement learning.
  arXiv  Detail & Related papers  (2022-09-26T17:59:58Z)
• Vision Transformer with Convolutions Architecture Search [72.70461709267497]
  We propose an architecture search method-Vision Transformer with Convolutions Architecture Search (VTCAS) The high-performance backbone network searched by VTCAS introduces the desirable features of convolutional neural networks into the Transformer architecture. It enhances the robustness of the neural network for object recognition, especially in the low illumination indoor scene.
  arXiv  Detail & Related papers  (2022-03-20T02:59:51Z)
• Improving Parametric Neural Networks for High-Energy Physics (and Beyond) [0.0]
  We aim at deepening the understanding of Parametric Neural Network (pNN) networks in light of real-world usage. We propose an alternative parametrization scheme, resulting in a new parametrized neural network architecture: the AffinePNN. We extensively evaluate our models on the HEPMASS dataset, along its imbalanced version (called HEPMASS-IMB)
  arXiv  Detail & Related papers  (2022-02-01T14:18:43Z)
• Conditionally Parameterized, Discretization-Aware Neural Networks for Mesh-Based Modeling of Physical Systems [0.0]
  We generalize the idea of conditional parametrization -- using trainable functions of input parameters. We show that conditionally parameterized networks provide superior performance compared to their traditional counterparts. A network architecture named CP-GNet is also proposed as the first deep learning model capable of reacting standalone prediction of flows on meshes.
  arXiv  Detail & Related papers  (2021-09-15T20:21:13Z)
• Efficient Micro-Structured Weight Unification and Pruning for Neural Network Compression [56.83861738731913]
  Deep Neural Network (DNN) models are essential for practical applications, especially for resource limited devices. Previous unstructured or structured weight pruning methods can hardly truly accelerate inference. We propose a generalized weight unification framework at a hardware compatible micro-structured level to achieve high amount of compression and acceleration.
  arXiv  Detail & Related papers  (2021-06-15T17:22:59Z)
• Modeling from Features: a Mean-field Framework for Over-parameterized Deep Neural Networks [54.27962244835622]
  This paper proposes a new mean-field framework for over- parameterized deep neural networks (DNNs) In this framework, a DNN is represented by probability measures and functions over its features in the continuous limit. We illustrate the framework via the standard DNN and the Residual Network (Res-Net) architectures.
  arXiv  Detail & Related papers  (2020-07-03T01:37:16Z)

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.