We introduce the community exploration problem that has many real-world applications such as online advertising.

General Value Functions (GVFs) (Sutton et al., 2011) represent predictive knowledge in reinforcement learning. Each GVF computes the expected return for a given policy, based on a unique reward. Existing methods relying on fixed behavior policies or pre-collected data often face data efficiency issues when learning multiple GVFs in parallel using off-policy methods. To address this, we introduce GVFExplorer, which adaptively learns a single behavior policy that efficiently collects data for evaluating multiple GVFs in parallel. We use an existing temporal-difference-style variance estimator to approximate the return variance.

Proximal Policy Optimization with Adaptive Exploration (axPPO) is introduced as a novel learning algorithm. This paper investigates the exploration-exploitation tradeoff within the context of reinforcement learning and aims to contribute new insights into reinforcement learning algorithm design. The proposed adaptive exploration framework dynamically adjusts the exploration magnitude during training based on the recent performance of the agent. Our proposed method outperforms standard PPO algorithms in learning efficiency, particularly when significant exploratory behavior is needed at the beginning of the learning process.

In multi-goal Reinforcement Learning, an agent can share experience between related training tasks, resulting in better generalization for new tasks at test time. However, when the goal space has discontinuities and the reward is sparse, a majority of goals are difficult to reach. In this context, a curriculum over goals helps agents learn by adapting training tasks to their current capabilities. In this work we propose Stein Variational Goal Generation (SVGG), which samples goals of intermediate difficulty for the agent, by leveraging a learned predictive model of its goal reaching capabilities. The distribution of goals is modeled with particles that are attracted in areas of appropriate difficulty using Stein Variational Gradient Descent. We show that SVGG outperforms state-of-the-art multi-goal Reinforcement Learning methods in terms of success coverage in hard exploration problems, and demonstrate that it is endowed with a useful recovery property when the environment changes.

Systems that make sequential decisions in the presence of partial feedback on actions often need to strike a balance between maximizing immediate payoffs based on available information, and acquiring new information that may be essential for maximizing future payoffs. This trade-off is captured by the multi-armed bandit (MAB) framework that has been studied and applied for designing sequential experiments when at each time epoch a single observation is collected on the action that was selected at that epoch. However, in many practical settings additional information may become available between decision epochs. We introduce a generalized MAB formulation in which auxiliary information on each arm may appear arbitrarily over time. By obtaining matching lower and upper bounds, we characterize the minimax complexity of this family of MAB problems as a function of the information arrival process, and study how salient characteristics of this process impact policy design and achievable performance. We establish the robustness of a Thompson sampling policy in the presence of additional information, but observe that other policies that are of practical importance do not exhibit such robustness. We therefore introduce a broad adaptive exploration approach for designing policies that, without any prior knowledge on the information arrival process, attain the best performance (in terms of regret rate) that is achievable when the information arrival process is a priori known. Our approach is based on adjusting MAB policies designed to perform well in the absence of auxiliary information by using dynamically customized virtual time indexes to endogenously control the exploration rate of the policy. We demonstrate our approach through appropriately adjusting known MAB policies and establishing improved performance bounds for these policies in the presence of auxiliary information.

We introduce the community exploration problem that has various real-world applications such as online advertising. In the problem, an explorer allocates limited budget to explore communities so as to maximize the number of members he could meet. We provide a systematic study of the community exploration problem, from offline optimization to online learning. For the offline setting where the sizes of communities are known, we prove that the greedy methods for both of non-adaptive exploration and adaptive exploration are optimal. For the online setting where the sizes of communities are not known and need to be learned from the multi-round explorations, we propose an ``upper confidence'' like algorithm that achieves the logarithmic regret bounds. By combining the feedback from different rounds, we can achieve a constant regret bound.