Two-Phase Dynamics of Interactions Explains the Starting Point of a DNN Learning Over-Fitted Features
- URL: http://arxiv.org/abs/2405.10262v1
- Date: Thu, 16 May 2024 17:13:25 GMT
- Title: Two-Phase Dynamics of Interactions Explains the Starting Point of a DNN Learning Over-Fitted Features
- Authors: Junpeng Zhang, Qing Li, Liang Lin, Quanshi Zhang,
- Abstract summary: We investigate the dynamics of a deep neural network (DNN) learning interactions.
In this paper, we discover the DNN learns interactions in two phases.
The first phase mainly penalizes interactions of medium and high orders, and the second phase mainly learns interactions of gradually increasing orders.
- Score: 68.3512123520931
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: This paper investigates the dynamics of a deep neural network (DNN) learning interactions. Previous studies have discovered and mathematically proven that given each input sample, a well-trained DNN usually only encodes a small number of interactions (non-linear relationships) between input variables in the sample. A series of theorems have been derived to prove that we can consider the DNN's inference equivalent to using these interactions as primitive patterns for inference. In this paper, we discover the DNN learns interactions in two phases. The first phase mainly penalizes interactions of medium and high orders, and the second phase mainly learns interactions of gradually increasing orders. We can consider the two-phase phenomenon as the starting point of a DNN learning over-fitted features. Such a phenomenon has been widely shared by DNNs with various architectures trained for different tasks. Therefore, the discovery of the two-phase dynamics provides a detailed mechanism for how a DNN gradually learns different inference patterns (interactions). In particular, we have also verified the claim that high-order interactions have weaker generalization power than low-order interactions. Thus, the discovered two-phase dynamics also explains how the generalization power of a DNN changes during the training process.
Related papers
- Defining and Extracting generalizable interaction primitives from DNNs [24.897875218177266]
We develop a new method to extract interactions that are shared by different deep neural networks (DNNs)
Experiments show that the extracted interactions can better reflect common knowledge shared by different DNNs.
arXiv Detail & Related papers (2024-01-29T17:21:41Z) - SEGNO: Generalizing Equivariant Graph Neural Networks with Physical
Inductive Biases [66.61789780666727]
We show how the second-order continuity can be incorporated into GNNs while maintaining the equivariant property.
We also offer theoretical insights into SEGNO, highlighting that it can learn a unique trajectory between adjacent states.
Our model yields a significant improvement over the state-of-the-art baselines.
arXiv Detail & Related papers (2023-08-25T07:15:58Z) - Technical Note: Defining and Quantifying AND-OR Interactions for
Faithful and Concise Explanation of DNNs [35.149004732692134]
We aim to explain a deep neural network (DNN) by quantifying the encoded interactions between input variables.
Specifically, we first rethink the definition of interactions, and then formally define faithfulness and conciseness for interaction-based explanation.
arXiv Detail & Related papers (2023-04-26T06:33:31Z) - Deep Architecture Connectivity Matters for Its Convergence: A
Fine-Grained Analysis [94.64007376939735]
We theoretically characterize the impact of connectivity patterns on the convergence of deep neural networks (DNNs) under gradient descent training.
We show that by a simple filtration on "unpromising" connectivity patterns, we can trim down the number of models to evaluate.
arXiv Detail & Related papers (2022-05-11T17:43:54Z) - Multi-scale Feature Learning Dynamics: Insights for Double Descent [71.91871020059857]
We study the phenomenon of "double descent" of the generalization error.
We find that double descent can be attributed to distinct features being learned at different scales.
arXiv Detail & Related papers (2021-12-06T18:17:08Z) - Discovering and Explaining the Representation Bottleneck of DNNs [21.121270460158712]
This paper explores the bottleneck of feature representations of deep neural networks (DNNs)
We focus on the multi-order interaction between input variables, where the order represents the complexity of interactions.
We discover that a DNN is more likely to encode both too simple interactions and too complex interactions, but usually fails to learn interactions of intermediate complexity.
arXiv Detail & Related papers (2021-11-11T14:35:20Z) - Interpreting Multivariate Shapley Interactions in DNNs [33.67263820904767]
This paper aims to explain deep neural networks (DNNs) from the perspective of multivariate interactions.
In this paper, we define and quantify the significance of interactions among multiple input variables of the DNN.
arXiv Detail & Related papers (2020-10-10T17:02:51Z) - Neural Additive Models: Interpretable Machine Learning with Neural Nets [77.66871378302774]
Deep neural networks (DNNs) are powerful black-box predictors that have achieved impressive performance on a wide variety of tasks.
We propose Neural Additive Models (NAMs) which combine some of the expressivity of DNNs with the inherent intelligibility of generalized additive models.
NAMs learn a linear combination of neural networks that each attend to a single input feature.
arXiv Detail & Related papers (2020-04-29T01:28:32Z) - Architecture Disentanglement for Deep Neural Networks [174.16176919145377]
We introduce neural architecture disentanglement (NAD) to explain the inner workings of deep neural networks (DNNs)
NAD learns to disentangle a pre-trained DNN into sub-architectures according to independent tasks, forming information flows that describe the inference processes.
Results show that misclassified images have a high probability of being assigned to task sub-architectures similar to the correct ones.
arXiv Detail & Related papers (2020-03-30T08:34:33Z)
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.