Interactive algorithms: from pool to stream
This work addresses the efficiency gap between pool and stream settings in interactive algorithms, which is incremental as it builds on existing interactive algorithm frameworks.
The paper tackles the problem of emulating pool-based interactive algorithms with stream-based ones, providing algorithms and matching lower bounds for general and utility-based pool algorithms, and demonstrates a maximal gap between the two settings in active learning for binary classification.
We consider interactive algorithms in the pool-based setting, and in the stream-based setting. Interactive algorithms observe suggested elements (representing actions or queries), and interactively select some of them and receive responses. Pool-based algorithms can select elements at any order, while stream-based algorithms observe elements in sequence, and can only select elements immediately after observing them. We assume that the suggested elements are generated independently from some source distribution, and ask what is the stream size required for emulating a pool algorithm with a given pool size. We provide algorithms and matching lower bounds for general pool algorithms, and for utility-based pool algorithms. We further show that a maximal gap between the two settings exists also in the special case of active learning for binary classification.