Mechanistic Mode Connectivity
Comments ICML, 2023
期刊&会议
ICML
International Conference on Machine Learning · 会议 · Machine Learning
Arithmetic Sampling: Parallel Diverse Decoding for Large Language Models
Comments 17 pages, to appear at ICML 2023
On the Identifiability and Estimation of Causal Location-Scale Noise Models
Comments ICML 2023
What Can Be Learnt With Wide Convolutional Neural Networks?
Journal ref Proceedings of the 40th International Conference on Machine Learning, PMLR 202. 2023
On Preemption and Learning in Stochastic Scheduling
Comments Accepted to ICML 2023
Graph Switching Dynamical Systems
Comments ICML 2023
Regularizing Towards Soft Equivariance Under Mixed Symmetries
Comments Proceedings of the International Conference on Machine Learning (ICML), 2023
Last Switch Dependent Bandits with Monotone Payoff Functions
Comments Accepted to the 40th International Conference on Machine Learning (ICML 2023)
Combinatorial Neural Bandits
Comments Accepted in ICML 2023
Conditionally Strongly Log-Concave Generative Models
Comments 28 pages, 12 figures, accepted at ICML 2023
Learning for Edge-Weighted Online Bipartite Matching with Robustness Guarantees
Comments Accepted by ICML 2023
Surrogate Model Extension (SME): A Fast and Accurate Weight Update Attack on Federated Learning
Comments Accepted at ICML 2023
Lottery Tickets in Evolutionary Optimization: On Sparse Backpropagation-Free Trainability
Comments 13 pages, 11 figures, International Conference on Machine Learning (ICML) 2023
Symmetry-Aware Robot Design with Structured Subgroups
Comments The Fortieth International Conference on Machine Learning (ICML 2023)
GraphCleaner: Detecting Mislabelled Samples in Popular Graph Learning Benchmarks
Comments ICML 2023
Crafting Training Degradation Distribution for the Accuracy-Generalization Trade-off in Real-World Super-Resolution
Comments This paper has been accepted to ICML 2023
The Implicit Regularization of Dynamical Stability in Stochastic Gradient Descent
Comments ICML 2023 camera ready
Data-OOB: Out-of-bag Estimate as a Simple and Efficient Data Value
Comments 18 pages, to be published at ICML 2023
When does Privileged Information Explain Away Label Noise?
Comments Accepted ICML 2023, Honolulu
Improved Algorithms for Multi-period Multi-class Packing Problems with Bandit Feedback
Comments Accepted in ICML 2023, 44 pages including Appendix
Equivariant Architectures for Learning in Deep Weight Spaces
Comments ICML 2023
Unearthing InSights into Mars: Unsupervised Source Separation with Limited Data
Comments ICML 2023
Graph Neural Tangent Kernel: Convergence on Large Graphs
Comments Accepted in the 40th International Conference on Machine Learning (ICML), Honolulu, Hawaii
Optimal randomized multilevel Monte Carlo for repeatedly nested expectations
Comments Accepted by ICML 2023. This version generalizes Thm 2.2 and 2.4 to multivariate underlying process, adds two numerical experiments, adds several references, and corrects several typos
Decentralized Stochastic Bilevel Optimization with Improved per-Iteration Complexity
Comments ICML 2023
From Perception to Programs: Regularize, Overparameterize, and Amortize
Comments ICML 2023
Improving Adversarial Robustness by Putting More Regularizations on Less Robust Samples
Comments Accepted in ICML 2023
A Law of Robustness beyond Isoperimetry
Comments To appear in ICML 2023
Monotonic Location Attention for Length Generalization
Comments Accepted in ICML 2023
MetaDiffuser: Diffusion Model as Conditional Planner for Offline Meta-RL
Comments 19 pages, 4 figures, accepted by ICML 23'
