Bayesian Neural Networks Avoid Encoding Complex and Perturbation-Sensitive Concepts

• URL: http://arxiv.org/abs/2302.13095v2
• Date: Fri, 1 Dec 2023 12:33:20 GMT
• Title: Bayesian Neural Networks Avoid Encoding Complex and Perturbation-Sensitive Concepts
• Authors: Qihan Ren, Huiqi Deng, Yunuo Chen, Siyu Lou, Quanshi Zhang
• Abstract summary: In this paper, we focus on mean-field variational Bayesian Neural Networks (BNNs) and explore the representation capacity of such BNNs. It has been observed and studied that a relatively small set of interactive concepts usually emerge in the knowledge representation of a sufficiently-trained neural network. Our study proves that compared to standard deep neural networks (DNNs), it is less likely for BNNs to encode complex concepts.
• Score: 22.873523599349326
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: In this paper, we focus on mean-field variational Bayesian Neural Networks (BNNs) and explore the representation capacity of such BNNs by investigating which types of concepts are less likely to be encoded by the BNN. It has been observed and studied that a relatively small set of interactive concepts usually emerge in the knowledge representation of a sufficiently-trained neural network, and such concepts can faithfully explain the network output. Based on this, our study proves that compared to standard deep neural networks (DNNs), it is less likely for BNNs to encode complex concepts. Experiments verify our theoretical proofs. Note that the tendency to encode less complex concepts does not necessarily imply weak representation power, considering that complex concepts exhibit low generalization power and high adversarial vulnerability. The code is available at https://github.com/sjtu-xai-lab/BNN-concepts.

Related papers

• Understanding polysemanticity in neural networks through coding theory [0.8702432681310401]
  We propose a novel practical approach to network interpretability and theoretical insights into polysemanticity and the density of codes. We show how random projections can reveal whether a network exhibits a smooth or non-differentiable code and hence how interpretable the code is. Our approach advances the pursuit of interpretability in neural networks, providing insights into their underlying structure and suggesting new avenues for circuit-level interpretability.
  arXiv  Detail & Related papers  (2024-01-31T16:31:54Z)
• A Primer on Bayesian Neural Networks: Review and Debates [10.074350562162921]
  Bayesian neural networks (BNNs) have emerged as a compelling extension of conventional neural networks. This primer presents a systematic introduction to the fundamental concepts of neural networks and Bayesian inference. The target audience comprises statisticians with a potential background in Bayesian methods but lacking deep learning expertise.
  arXiv  Detail & Related papers  (2023-09-28T10:09:15Z)
• Explaining Generalization Power of a DNN Using Interactive Concepts [24.712192363947096]
  This paper explains the generalization power of a deep neural network (DNN) from the perspective of interactions. We also discover the detouring dynamics of learning complex concepts, which explains both the high learning difficulty and the low generalization power of complex concepts.
  arXiv  Detail & Related papers  (2023-02-25T14:44:40Z)
• Does a Neural Network Really Encode Symbolic Concepts? [24.099892982101398]
  In this paper, we examine the trustworthiness of interaction concepts from four perspectives. Extensive empirical studies have verified that a well-trained DNN usually encodes sparse, transferable, and discriminative concepts.
  arXiv  Detail & Related papers  (2023-02-25T13:58:37Z)
• Basic Binary Convolution Unit for Binarized Image Restoration Network [146.0988597062618]
  In this study, we reconsider components in binary convolution, such as residual connection, BatchNorm, activation function, and structure, for image restoration tasks. Based on our findings and analyses, we design a simple yet efficient basic binary convolution unit (BBCU) Our BBCU significantly outperforms other BNNs and lightweight models, which shows that BBCU can serve as a basic unit for binarized IR networks.
  arXiv  Detail & Related papers  (2022-10-02T01:54:40Z)
• Robustness of Bayesian Neural Networks to White-Box Adversarial Attacks [55.531896312724555]
  Bayesian Networks (BNNs) are robust and adept at handling adversarial attacks by incorporating randomness. We create our BNN model, called BNN-DenseNet, by fusing Bayesian inference (i.e., variational Bayes) to the DenseNet architecture. An adversarially-trained BNN outperforms its non-Bayesian, adversarially-trained counterpart in most experiments.
  arXiv  Detail & Related papers  (2021-11-16T16:14:44Z)
• BCNN: Binary Complex Neural Network [16.82755328827758]
  Binarized neural networks, or BNNs, show great promise in edge-side applications with resource limited hardware. We introduce complex representation into the BNNs and propose Binary complex neural network. BCNN improves BNN by strengthening its learning capability through complex representation and extending its applicability to complex-valued input data.
  arXiv  Detail & Related papers  (2021-03-28T03:35:24Z)
• S2-BNN: Bridging the Gap Between Self-Supervised Real and 1-bit Neural Networks via Guided Distribution Calibration [74.5509794733707]
  We present a novel guided learning paradigm from real-valued to distill binary networks on the final prediction distribution. Our proposed method can boost the simple contrastive learning baseline by an absolute gain of 5.515% on BNNs. Our method achieves substantial improvement over the simple contrastive learning baseline, and is even comparable to many mainstream supervised BNN methods.
  arXiv  Detail & Related papers  (2021-02-17T18:59:28Z)
• Neural Networks Enhancement with Logical Knowledge [83.9217787335878]
  We propose an extension of KENN for relational data. The results show that KENN is capable of increasing the performances of the underlying neural network even in the presence relational data.
  arXiv  Detail & Related papers  (2020-09-13T21:12:20Z)
• Neural Networks with Recurrent Generative Feedback [61.90658210112138]
  We instantiate this design on convolutional neural networks (CNNs) In the experiments, CNN-F shows considerably improved adversarial robustness over conventional feedforward CNNs on standard benchmarks.
  arXiv  Detail & Related papers  (2020-07-17T19:32:48Z)
• A Chain Graph Interpretation of Real-World Neural Networks [58.78692706974121]
  We propose an alternative interpretation that identifies NNs as chain graphs (CGs) and feed-forward as an approximate inference procedure. The CG interpretation specifies the nature of each NN component within the rich theoretical framework of probabilistic graphical models. We demonstrate with concrete examples that the CG interpretation can provide novel theoretical support and insights for various NN techniques.
  arXiv  Detail & Related papers  (2020-06-30T14:46:08Z)

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.