NeurIPS 2024PastAI for science
NeurIPS 2024 Workshop on Scientific Methods for Understanding Deep Learning
SciForDL
- Submission deadline
- Sep 18, 2024, 12:59 UTCimported from OpenReview — check the website for extensions
- Submission portal
- OpenReview
- Notes
- Topics were auto-suggested and may be imprecise — edits welcome.
Accepted papers (73)
Fetched from OpenReview (v2) on 2026-06-10.
A Continuous-Time Analysis of Adaptive Optimization and Normalization
A Method on Searching Better Activation Functions
Alice in Wonderland: Simple Tasks Reveal Severe Generalization and Basic Reasoning Deficits in State-Of-the-Art Large Language Models
Amplified Early Stopping Bias: Overestimated Performance with Deep Learning
Are Capsule Networks Texture or Shape Biased?
BatchTopK Sparse Autoencoders
Causation Does Not Imply Correlation: A Study of Circuit Mechanisms and Model Behaviors
Characterizing stable regions in the residual stream of LLMs
Comparing Apples and Oranges: is Stitching Similarity a Load of Spheres?
Denoising for Manifold Extrapolation
Distributional Scaling Laws for Emergent Capabilities
Effectiveness of Sparse Autoencoder for understanding and removing gender bias in LLMs
Eliminating Position Bias of Language Models: A Mechanistic Approach
Emergence of Hierarchical Emotion Representations in Large Language Models
Emergent properties with repeated examples
EmoCAM: Toward Understanding What Drives CNN-based Emotion Recognition
Evaluating Loss Landscapes from a Topology Perspective
Explicit Regularisation, Sharpness and Calibration
Exploiting Interpretable Capabilities with Concept-Enhanced Diffusion and Prototype Networks
Exploring model depth and data complexity through the lens of cellular automata
Generalization vs Specialization under Concept Shift
Hiding in a Plain Sight: Out-of-Distribution Data in the Logit Space Embeddings
How Learning Rates Shape Neural Network Focus: Insights from Example Ranking
How rare events shape the learning curves of hierarchical data
Illusions as features: the generative side of recognition
Impact of Label Noise on Learning Complex Features
Improving Deep Learning Speed and Performance through Synaptic Neural Balance
Input Space Mode Connectivity in Deep Neural Networks
Interpretability as Compression: Reconsidering SAE Explanations of Neural Activations
Investigating Sensitive Directions in GPT-2: An Improved Baseline and Comparative Analysis of SAEs
Is Expressivity Essential for the Predictive Performance of Graph Neural Networks?
Is network fragmentation a useful complexity measure?
Is Saliency Really Captured By Gradient?
Knowledge Distillation for Teaching Symmetry Invariances
Knowledge Distillation: The Functional Perspective
Language model scaling laws and zero-sum learning
Learnability in the Context of Neural Tangent Kernels
Learned Random Label Predictions as a Neural Network Complexity Metric
Learning Stochastic Rainbow Networks
Logicbreaks: A Framework for Understanding Subversion of Rule-based Inference
Memorization to Generalization: The Emergence of Diffusion Models from Associative Memory
Model Recycling: Model component reuse to promote in-context learning
On the Collapse Errors Induced by the Deterministic Sampler for Diffusion Models
Pre-processing and Compression: Understanding Hidden Representation Refinement Across Imaging Domains via Intrinsic Dimension
Probing the Decision Boundaries of In-context Learning in Large Language Models Download PDF
Rethinking Knowledge Transfer in Learning Using Privileged Information
Revealing the Learning Process in Reinforcement Learning Agents Through Attention-Oriented Metrics
Robust Learning in Bayesian Parallel Branching Graph Neural Networks: The Narrow Width Limit
softmax is not enough (for sharp out-of-distribution)
SolidMark: How to Evaluate Memorization in Image Generative Models
Sometimes I am a Tree: Data Drives Fragile Hierarchical Generalization
Sparse autoencoders for dense text embeddings reveal hierarchical feature sub-structure
Specialization-generalization transition in exemplar-based in-context learning
Standard adversarial attacks only fool the final layer
Stitching Sparse Autoencoders of Different Sizes
Structure Development in List Sorting Transformers
Structured Identity Mapping Learning As a Model for Compositional Generalization in Generative Models
Testing knowledge distillation theories with dataset size
The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains
The Master Key Filters Hypothesis: Deep Filters Are General
The Pitfalls of Memorization: When Memorization Hinders Generalization
The Unreasonable Ineffectiveness of the Deeper Layers
Token-token correlations predict the scaling of the test loss with the number of input tokens
Towards Understanding In-Context Learning with Contrastive Demonstrations and Saliency Maps
Training Dynamics of Convolutional Neural Networks for Learning the Derivative Operator
Training Neural Networks for Modularity aids Interpretability
Transformers can reinforcement learn to approximate Gittins Index
Twin Studies of Factors in OOD Generalization
Understanding the Limitations of B-Spline KANs: Convergence Dynamics and Computational Efficiency
Understanding the Transient Nature of In-Context Learning: The Window of Generalization
Understanding Visual Concepts Across Models
Unraveling the Latent Hierarchical Structure of Language and Images via Diffusion Models
We Need Far Fewer Unique Filters Than We Thought