Machine Learning via Sequential Interaction (CS-0128)

Spring 2026 · CS 0128 · Tufts University · 3 credits · Undergraduate

Introduction of machine learning as a sequential and interactive game technically referred to as 'online learning'. In online learning, the learner aims to minimize the regret: loss incurred for not playing the best strategy (in hindsight) averaged over time. This framework generalizes the traditional statistical learning set-up in a non-trivial manner to cases where the examples are not independent, are supplied only sequentially, and possibly in an adversarial manner. Topics include: general set-up of sequential learning with examples, online convex and online linear optimization, online-to-batch conversion, multi-armed bandits, contextual bandits, practical algorithms and their analysis with provable guarantees. Connections with and extensions to reinforcement learning will be highlighted. Prior programming experience is recommended.

Course codes: CS-0128, CS 0128, CS0128, CS-128, CS 128, CS128, EE-0140, EE 0140, EE0140, EE-140, EE 140, EE140

Course Discovery