Similarity of Neural Network Representations Revisited
arXiv:1905.00414
Abstract
Recent work has sought to understand the behavior of neural networks by comparing representations between layers and between different trained models. We examine methods for comparing neural network representations based on canonical correlation analysis (CCA). We show that CCA belongs to a family of statistics for measuring multivariate similarity, but that neither CCA nor any other statistic that is invariant to invertible linear transformation can measure meaningful similarities between representations of higher dimension than the number of data points. We introduce a similarity index that measures the relationship between representational similarity matrices and does not suffer from this limitation. This similarity index is equivalent to centered kernel alignment (CKA) and is also closely connected to CCA. Unlike CCA, CKA can reliably identify correspondences between representations in networks trained from different initializations.
arXiv commentICML 2019
Cited by in corpus (119)
- Transfusion: Understanding Transfer Learning for Medical Imaging
- Convolutional Neural Networks as a Model of the Visual System: Past, Present, and Future
- Artificial neural networks for neuroscientists: A primer
- What is being transferred in transfer learning?
- Do Vision Transformers See Like Convolutional Neural Networks?
- Do Wide and Deep Networks Learn the Same Things? Uncovering How Neural Network Representations Vary with Width and Depth
- Rapid Learning or Feature Reuse? Towards Understanding the Effectiveness of MAML
- Overlearning Reveals Sensitive Attributes
- On Representation Knowledge Distillation for Graph Neural Networks
- How deep is deep enough? -- Quantifying class separability in the hidden layers of deep neural networks
- BOIL: Towards Representation Change for Few-shot Learning
- A Survey of Deep Learning for Scientific Discovery
- What shapes feature representations? Exploring datasets, architectures, and training
- Adaptive Model Pooling for Online Deep Anomaly Detection from a Complex Evolving Data Stream
- Self-Supervised Pre-Training for Transformer-Based Person Re-Identification
- Learning Student-Friendly Teacher Networks for Knowledge Distillation
- Entangled Watermarks as a Defense against Model Extraction
- Advancing diagnostic performance and clinical usability of neural networks via adversarial training and dual batch normalization
- No Fear of Heterogeneity: Classifier Calibration for Federated Learning with Non-IID Data
- Universality and individuality in neural dynamics across large populations of recurrent networks
- Universal Paralinguistic Speech Representations Using Self-Supervised Conformers
- Quantifying the LiDAR Sim-to-Real Domain Shift: A Detailed Investigation Using Object Detectors and Analyzing Point Clouds at Target-Level
- LGViT: Dynamic Early Exiting for Accelerating Vision Transformer
- Emerging Cross-lingual Structure in Pretrained Language Models
- The Shape of Data: Intrinsic Distance for Data Distributions
- Low-Dimensional Structure in the Space of Language Representations is Reflected in Brain Responses
- Supervised Transfer Learning at Scale for Medical Imaging
- Revisiting Model Stitching to Compare Neural Representations
- Radio2Text: Streaming Speech Recognition Using mmWave Radio Signals
- Knowledge Consistency between Neural Networks and Beyond
- Unraveling Meta-Learning: Understanding Feature Representations for Few-Shot Tasks
- Object Detector Differences when using Synthetic and Real Training Data
- Explaining Knowledge Distillation by Quantifying the Knowledge
- The intriguing role of module criticality in the generalization of deep networks
- Similarity of Neural Networks with Gradients
- The effect of task and training on intermediate representations in convolutional neural networks revealed with modified RV similarity analysis
- Hierarchically Organized Latent Modules for Exploratory Search in Morphogenetic Systems
- Linear Mode Connectivity in Multitask and Continual Learning
- The Role of Complex NLP in Transformers for Text Ranking?
- An Information Theory-inspired Strategy for Automatic Network Pruning
- Machine Learning for Cataract Classification and Grading on Ophthalmic Imaging Modalities: A Survey
- Context Meta-Reinforcement Learning via Neuromodulation
- Probing the State of the Art: A Critical Look at Visual Representation Evaluation
- Taxonomizing local versus global structure in neural network loss landscapes
- Is Supervised Syntactic Parsing Beneficial for Language Understanding? An Empirical Investigation
- OMPQ: Orthogonal Mixed Precision Quantization
- An Empirical Study on Post-processing Methods for Word Embeddings
- PURSUhInT: In Search of Informative Hint Points Based on Layer Clustering for Knowledge Distillation
- TRS: Transferability Reduced Ensemble via Encouraging Gradient Diversity and Model Smoothness
- Teachers Do More Than Teach: Compressing Image-to-Image Models
- Insights on Neural Representations for End-to-End Speech Recognition
- On Negative Interference in Multilingual Models: Findings and A Meta-Learning Treatment
- Nondeterminism and Instability in Neural Network Optimization
- VendorLink: An NLP approach for Identifying & Linking Vendor Migrants & Potential Aliases on Darknet Markets
- Transferred Discrepancy: Quantifying the Difference Between Representations
- SoK: How Robust is Image Classification Deep Neural Network Watermarking? (Extended Version)
- Exploring the Interchangeability of CNN Embedding Spaces
- Harmonizing Feature Attributions Across Deep Learning Architectures: Enhancing Interpretability and Consistency
- Simon Says: Evaluating and Mitigating Bias in Pruned Neural Networks with Knowledge Distillation
- Cross-Layer Distillation with Semantic Calibration
- Grounding Representation Similarity with Statistical Testing
- Exploring Linguistic Properties of Monolingual BERTs with Typological Classification among Languages
- A Channel Coding Benchmark for Meta-Learning
- FEAR: A Simple Lightweight Method to Rank Architectures
- ACDC: Weight Sharing in Atom-Coefficient Decomposed Convolution
- When to Prune? A Policy towards Early Structural Pruning
- Invariance Measures for Neural Networks
- Robust Distance Covariance
- Compare Where It Matters: Using Layer-Wise Regularization To Improve Federated Learning on Heterogeneous Data
- Graph-Based Similarity of Neural Network Representations
- Undivided Attention: Are Intermediate Layers Necessary for BERT?
- Representation Transfer by Optimal Transport
- Knowledge Distillation Meets Self-Supervision
- Net2Brain: A Toolbox to compare artificial vision models with human brain responses
- Revisiting Hidden Representations in Transfer Learning for Medical Imaging
- Phone and speaker spatial organization in self-supervised speech representations
- Aggregative Self-Supervised Feature Learning from a Limited Sample
- Learning Compact Representations of Neural Networks using DiscriminAtive Masking (DAM)
- Interpreting and Disentangling Feature Components of Various Complexity from DNNs
- NomMer: Nominate Synergistic Context in Vision Transformer for Visual Recognition
- Leveraging Undiagnosed Data for Glaucoma Classification with Teacher-Student Learning
- Hierarchical nucleation in deep neural networks
- Do Self-Supervised and Supervised Methods Learn Similar Visual Representations?
- Exploring the limits of pre-trained embeddings in machine-guided protein design: a case study on predicting AAV vector viability
- Differential Privacy, Linguistic Fairness, and Training Data Influence: Impossibility and Possibility Theorems for Multilingual Language Models
- Properties of the After Kernel
- Trivial or impossible -- dichotomous data difficulty masks model differences (on ImageNet and beyond)
- How do Quadratic Regularizers Prevent Catastrophic Forgetting: The Role of Interpolation
- FedProf: Selective Federated Learning with Representation Profiling
- Comparison Against Task Driven Artificial Neural Networks Reveals Functional Organization of Mouse Visual Cortex
- Semi-Online Knowledge Distillation
- Improved Knowledge Distillation via Adversarial Collaboration
- Multi-level Knowledge Distillation via Knowledge Alignment and Correlation
- Why Do Better Loss Functions Lead to Less Transferable Features?
- Identifying the Correlation Between Language Distance and Cross-Lingual Transfer in a Multilingual Representation Space
- Similarity Analysis of Self-Supervised Speech Representations
- Enhancing Data-Free Adversarial Distillation with Activation Regularization and Virtual Interpolation
- ImageNet Pre-training also Transfers Non-Robustness
- Pull-back Geometry of Persistent Homology Encodings
- An Investigation of the Weight Space to Monitor the Training Progress of Neural Networks
- Multi-granularity for knowledge distillation
- How Familiar Does That Sound? Cross-Lingual Representational Similarity Analysis of Acoustic Word Embeddings
- Low Frequency Names Exhibit Bias and Overfitting in Contextualizing Language Models
- LOOPER: Inferring computational algorithms enacted by neuronal population dynamics
- Characterizing and Measuring the Similarity of Neural Networks with Persistent Homology
- Meta-Learning to Improve Pre-Training
- Exploiting a Zoo of Checkpoints for Unseen Tasks
- Does the Adam Optimizer Exacerbate Catastrophic Forgetting?
- Verifiability and Predictability: Interpreting Utilities of Network Architectures for Point Cloud Processing
- Facial Landmark Correlation Analysis
- Disentangling Transfer and Interference in Multi-Domain Learning
- Back to Square One: Superhuman Performance in Chutes and Ladders Through Deep Neural Networks and Tree Search
- Understanding the Dynamics of DNNs Using Graph Modularity
- On Cross-Layer Alignment for Model Fusion of Heterogeneous Neural Networks
- One-Shot Neural Ensemble Architecture Search by Diversity-Guided Search Space Shrinking
- OrthoReg: Robust Network Pruning Using Orthonormality Regularization
- Correlations between Word Vector Sets
- A Methodology for Exploring Deep Convolutional Features in Relation to Hand-Crafted Features with an Application to Music Audio Modeling
- Reducing the Human Effort in Developing PET-CT Registration