A Compression Perspective on Simplicity Bias
Deep neural networks exhibit a simplicity bias, a well-documented tendency to favor simple functions over complex ones. In this work, we cast new light on this phenomenon through the lens of the Minimum Description Length principle, formalizing supervised learning as a problem of optimal two-part lossless compression. Our theory explains how simplicity bias governs feature selection in neural networks through a fundamental trade-off between model complexity (the cost of describing the hypothesis) and predictive power (the cost of describing the data). Our framework predicts that as the amount of available training data increases, learners transition through qualitatively different features -- from simple spurious shortcuts to complex features -- only when the reduction in data encoding cost justifies the increased model complexity. Consequently, we identify distinct data regimes where increasing data promotes robustness by ruling out trivial shortcuts, and conversely, regimes where limiting data can act as a form of complexity-based regularization, preventing the learning of unreliable complex environmental cues. We validate our theory on a semi-synthetic benchmark showing that the feature selection of neural networks follows the same trajectory of solutions as optimal two-part compressors.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity
We optimize rewards of reinforcement learning in text simplification using metrics that are highly correlated with human-perspectives. To address problems of exposure bias and loss-evaluation mismatch, text-to-text gener…
reinforcement-learningReinforcement LearningReinforcement Learning (RL)Text Generation+1Simplicity bias, algorithmic probability, and the random logistic map
Simplicity bias is an intriguing phenomenon prevalent in various input-output maps, characterized by a preference for simpler, more regular, or symmetric outputs. Notably, these maps typically feature high-probability ou…
Understanding Simplicity Bias towards Compositional Mappings via Learning Dynamics
Obtaining compositional mappings is important for the model to generalize well compositionally. To better understand when and how to encourage the model to learn such mappings, we study their uniqueness through different…
Simplicity bias in the parameter-function map of deep neural networks
The idea that neural networks may exhibit a bias towards simplicity has a long history. Simplicity bias provides a way to quantify this intuition. It predicts, for a broad class of input-output maps which can describe m…
Simplicity Bias of Transformers to Learn Low Sensitivity Functions
Transformers achieve state-of-the-art accuracy and robustness across many tasks, but an understanding of the inductive biases that they have and how those biases are different from other neural network architectures rema…
Sensitivity