The Sample Complexity of Parameter-Free Stochastic Convex Optimization
By: Jared Lawrence , Ari Kalinsky , Hannah Bradfield and more
Potential Business Impact:
Teaches computers to learn better from less data.
We study the sample complexity of stochastic convex optimization when problem parameters, e.g., the distance to optimality, are unknown. We pursue two strategies. First, we develop a reliable model selection method that avoids overfitting the validation set. This method allows us to generically tune the learning rate of stochastic optimization methods to match the optimal known-parameter sample complexity up to $\log\log$ factors. Second, we develop a regularization-based method that is specialized to the case that only the distance to optimality is unknown. This method provides perfect adaptability to unknown distance to optimality, demonstrating a separation between the sample and computational complexity of parameter-free stochastic convex optimization. Combining these two methods allows us to simultaneously adapt to multiple problem structures. Experiments performing few-shot learning on CIFAR-10 by fine-tuning CLIP models and prompt engineering Gemini to count shapes indicate that our reliable model selection method can help mitigate overfitting to small validation sets.
Similar Papers
Convergence and Sample Complexity of First-Order Methods for Agnostic Reinforcement Learning
Machine Learning (CS)
Teaches computers to learn the best way to do things.
Optimality and computational barriers in variable selection under dependence
Statistics Theory
Finds the best way to pick important data.
Single-loop Algorithms for Stochastic Non-convex Optimization with Weakly-Convex Constraints
Machine Learning (CS)
Makes AI learn better with fewer steps.