# Emergence of Invariance and Disentanglement in Deep Representations

arXiv: https://arxiv.org/abs/1706.01350 · HTML: https://mlsanity.com/paper/1706.01350/

## Key facts

- Cited by 215 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 5 Jun 2017.

## Built on by (most cited first, 30 of 215)

- [Task Arithmetic in the Tangent Space: Improved Editing of Pre-Trained Models](https://mlsanity.com/paper/2305.12827/) (2023)
- [Rapid Exploration for Open-World Navigation with Latent Goal Models](https://mlsanity.com/paper/2104.05859/) (2021)
- [Uncovering the Disentanglement Capability in Text-to-Image Diffusion Models](https://mlsanity.com/paper/2212.08698/) (2022)
- [How Does Information Bottleneck Help Deep Learning?](https://mlsanity.com/paper/2305.18887/) (2023)
- [Exploiting Domain-Specific Features to Enhance Domain Generalization](https://mlsanity.com/paper/2110.09410/) (2021)
- [Graph Structure Learning with Variational Information Bottleneck](https://mlsanity.com/paper/2112.08903/) (2021)
- [Desiderata for Representation Learning: A Causal Perspective](https://mlsanity.com/paper/2109.03795/) (2021)
- [To Compress or Not to Compress- Self-Supervised Learning and Information Theory: A Review](https://mlsanity.com/paper/2304.09355/) (2023)
- [Revisiting Locally Supervised Learning: an Alternative to End-to-end Training](https://mlsanity.com/paper/2101.10832/) (2021)
- [A Workflow for Offline Model-Free Robotic Reinforcement Learning](https://mlsanity.com/paper/2109.10813/) (2021)
- [Rethinking Minimal Sufficient Representation in Contrastive Learning](https://mlsanity.com/paper/2203.07004/) (2022)
- [Examining and Combating Spurious Features under Distribution Shift](https://mlsanity.com/paper/2106.07171/) (2021)
- [Factorizing Knowledge in Neural Networks](https://mlsanity.com/paper/2207.03337/) (2022)
- [Deep Gait Recognition: A Survey](https://mlsanity.com/paper/2102.09546/) (2021)
- [Non-Transferable Learning: A New Approach for Model Ownership Verification and Applicability Authorization](https://mlsanity.com/paper/2106.06916/) (2021)
- [Linear Spaces of Meanings: Compositional Structures in Vision-Language Models](https://mlsanity.com/paper/2302.14383/) (2023)
- [Learning Disentangled Representations in the Imaging Domain](https://mlsanity.com/paper/2108.12043/) (2021)
- [Pre-training Contextualized World Models with In-the-wild Videos for Reinforcement Learning](https://mlsanity.com/paper/2305.18499/) (2023)
- [HYperbolic Self-Paced Learning for Self-Supervised Skeleton-based Action Representations](https://mlsanity.com/paper/2303.06242/) (2023)
- [Rethinking Confidence Calibration for Failure Prediction](https://mlsanity.com/paper/2303.02970/) (2023)
- [Compressive Visual Representations](https://mlsanity.com/paper/2109.12909/) (2021)
- [A Variational Information Bottleneck Approach to Multi-Omics Data Integration](https://mlsanity.com/paper/2102.03014/) (2021)
- [Differentiation and Specialization of Attention Heads via the Refined Local Learning Coefficient](https://mlsanity.com/paper/2410.02984/) (2024)
- [Why neural networks find simple solutions: the many regularizers of geometric complexity](https://mlsanity.com/paper/2209.13083/) (2022)
- [Shape-Erased Feature Learning for Visible-Infrared Person Re-Identification](https://mlsanity.com/paper/2304.04205/) (2023)
- [Efficient Knowledge Distillation from Model Checkpoints](https://mlsanity.com/paper/2210.06458/) (2022)
- [Semi-supervised Meta-learning with Disentanglement for Domain-generalised Medical Image Segmentation](https://mlsanity.com/paper/2106.13292/) (2021)
- [Probabilistic Representations for Video Contrastive Learning](https://mlsanity.com/paper/2204.03946/) (2022)
- [Robust Predictable Control](https://mlsanity.com/paper/2109.03214/) (2021)
- [Recognizing Predictive Substructures with Subgraph Information Bottleneck](https://mlsanity.com/paper/2103.11155/) (2021)

## Abstract (arXiv)

Using established principles from Statistics and Information Theory, we show that invariance to nuisance factors in a deep neural network is equivalent to information minimality of the learned representation, and that stacking layers and injecting noise during training naturally bias the network towards learning invariant representations. We then decompose the cross-entropy loss used during training and highlight the presence of an inherent overfitting term. We propose regularizing the loss by bounding such a term in two equivalent ways: One with a Kullbach-Leibler term, which relates to a PAC-Bayes perspective; the other using the information in the weights as a measure of complexity of a learned model, yielding a novel Information Bottleneck for the weights. Finally, we show that invariance and independence of the components of the representation learned by the network are bounded above and below by the information in the weights, and therefore are implicitly optimized during training. The theory enables us to quantify and predict sharp phase transitions between underfitting and overfitting of random labels when using our regularized loss, which we verify in experiments, and sheds light on the relation between the geometry of the loss function, invariance properties of the learned representation, and generalization error.