Shie Mannor

Overlapping Community Detection by Online Cluster Aggregation

2015 · arXiv (Cornell University)

We present a new online algorithm for detecting overlapping communities. The main ingredients are a modification of an online k-means algorithm and a new approach to modelling overlap in communities. An evaluation on large benchmark …

Contextual Markov Decision Processes

2015 · arXiv (Cornell University)

We consider a planning problem where the dynamics and rewards of the environment depend on a hidden static parameter referred to as the context. The objective is to learn a strategy that maximizes the accumulated …

How to Allocate Resources For Features Acquisition?

2016 · arXiv (Cornell University)

We study classification problems where features are corrupted by noise and where the magnitude of the noise in each feature is influenced by the resources allocated to its acquisition. This is the case, for example, …

Finite Sample Analyses for TD(0) with Function Approximation

2017 · arXiv (Cornell University)

TD(0) is one of the most commonly used algorithms in reinforcement learning. Despite this, there is no existing finite sample analysis for TD(0) with function approximation, even for the linear case. Our work is the …

Finite Sample Analysis of Two-Timescale Stochastic Approximation with Applications to Reinforcement Learning

2017 · arXiv (Cornell University)

Two-timescale Stochastic Approximation (SA) algorithms are widely used in Reinforcement Learning (RL). Their iterates have two parts that are updated using distinct stepsizes. In this work, we develop a novel recipe for their finite sample …

Bootstrapping Skills

2015 · arXiv (Cornell University)

The monolithic approach to policy representation in Markov Decision Processes (MDPs) looks for a single policy that can be represented as a function from states to actions. For the monolithic approach to succeed (and this …

Detecting Rewards Deterioration in Episodic Reinforcement Learning

2020 · arXiv (Cornell University)

In many RL applications, once training ends, it is vital to detect any deterioration in the agent performance as soon as possible. Furthermore, it often has to be done without modifying the policy and under …

Individualized Dosing Dynamics via Neural Eigen Decomposition

2023 · arXiv (Cornell University)

Dosing models often use differential equations to model biological dynamics. Neural differential equations in particular can learn to predict the derivative of a process, which permits predictions at irregular points of time. However, this temporal …