A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
Optimizes quantile and semi-adversarial regret with novel root-logarithmic regularizers.
Adapting Hedge algorithm for semi-adversarial data with root-entropy regularization.
We consider the problem of controlling a possibly unknown linear dynamical system with adversarial perturbations, adversarially chosen convex loss functions, and partially observed states, known as non-stochastic control. We introduce a controller parametrization based on the denoised observations, and prove that apply…
Algorithm learns expert weights to minimize regret in adversarial setting.
Algorithm learns without knowing distribution, reducing error.