Infrastructure for Contextual Bandits and Reinforcement Learning -- theme of the ML Platform meetup hosted at Netflix, Los Gatos on Sep 12, 2019. Contextual and Multi-armed Bandits enable faster and adaptive alternatives to traditional A/B Testing. They enable rapid learning and better decision-making for product rollouts. Broadly speaking, these approaches can be seen as a stepping stone to full-on Reinforcement Learning (RL) with closed-loop, on-policy evaluation and model objectives tied to reward functions. At Netflix, we are running several such experiments.

Personal use of this material is permitted. Abstract --In this work, we study semi-supervised multi-label node classification problem in attributed graphs. Classic solutions to multi-label node classification follow two steps, first learn node embedding and then build a node classifier on the learned embedding. T o improve the discriminating power of the node embedding, we propose a novel collaborative graph walk, named Multi-Label-Graph-Walk, to finely tune node representations with the available label assignments in attributed graphs via reinforcement learning. The proposed method formulates the multi-label node classification task as simultaneous graph walks conducted by multiple label-specific agents. Furthermore, policies of the label-wise graph walks are learned in a cooperative way to capture first the predictive relation between node labels and structural attributes of graphs; and second, the correlation among the multiple label-specific classification tasks. A comprehensive experimental study demonstrates that the proposed method can achieve significantly better multi-label classification performance than the state-of-the-art approaches and conduct more efficient graph exploration. Index T erms --Multi-label node classification, Semi-supervised attributed graph embedding, Reinforcement learning I. I NTRODUCTION Graph-structured data are frequently witnessed in many real-world applications, such as social graphs and academic graphs. In the graph structure, nodes represent entities (e.g., users in social graphs and papers in citation graphs), whereas edges linking two nodes denote the relationship between the entities (e.g., user friendship and paper citation). Usually both nodes and edges possess their own attributes.

In this paper, we study reinforcement learning (RL) algorithms to solve real-world decision problems with the objective of maximizing the long-term reward as well as satisfying cumulative constraints. We propose a novel first-order policy optimization method, Interior-point Policy Optimization (IPO), which augments the objective with logarithmic barrier functions, inspired by the interior-point method. Our proposed method is easy to implement with performance guarantees and can handle general types of cumulative multi-constraint settings. We conduct extensive evaluations to compare our approach with state-of-the-art baselines. Our algorithm outperforms the baseline algorithms, in terms of reward maximization and constraint satisfaction. Introduction Recent advances have demonstrated significant potentials of deep reinforcement learning (RL) in solving complex sequential decision and control problems, e.g., the Atari game (Mnih et al. 2015), robotics (Andrychowicz et al. 2018), Go (Silver et al. 2016), etc. In such RL problems, the objective is to maximize the discounted cumulative reward. In many other problems, in addition to maximizing the reward, a policy needs to satisfy certain constraints.

Language creates a compact representation of the world and allows the description of unlimited situations and objectives through compositionality. These properties make it a natural fit to guide the training of interactive agents as it could ease recurrent challenges in Reinforcement Learning such as sample complexity, generalization, or multi-tasking. Yet, it remains an open-problem to relate language and RL in even simple instruction following scenarios. Current methods rely on expert demonstrations, auxiliary losses, or inductive biases in neural architectures. In this paper, we propose an orthogonal approach called Textual Hindsight Experience Replay (THER) that extends the Hindsight Experience Replay approach to the language setting. Whenever the agent does not fulfill its instruction, THER learns to output a new directive that matches the agent trajectory, and it relabels the episode with a positive reward. To do so, THER learns to map a state into an instruction by using past successful trajectories, which removes the need to have external expert interventions to relabel episodes as in vanilla HER. We observe that this simple idea also initiates a learning synergy between language acquisition and policy learning on instruction following tasks in the BabyAI environment.

We adapt the optimization's concept of momentum to reinforcement learning. Seeing the state-action value functions as an analog to the gradients in optimization, we interpret momentum as an average of consecutive $q$-functions. We derive Momentum Value Iteration (MoVI), a variation of Value Iteration that incorporates this momentum idea. Our analysis shows that this allows MoVI to average errors over successive iterations. We show that the proposed approach can be readily extended to deep learning. Specifically, we propose a simple improvement on DQN based on MoVI, and experiment it on Atari games.

Deep reinforcement learning (DRL) is a booming area of artificial intelligence. Many practical applications of DRL naturally involve more than one collaborative learners, making it important to study DRL in a multi-agent context. Previous research showed that effective learning in complex multi-agent systems demands for highly coordinated environment exploration among all the participating agents. Many researchers attempted to cope with this challenge through learning centralized value functions. However, the common strategy for every agent to learn their local policies directly often fail to nurture strong inter-agent collaboration and can be sample inefficient whenever agents alter their communication channels. To address these issues, we propose a new framework known as centralized training and exploration with decentralized execution via policy distillation. Guided by this framework and the maximum-entropy learning technique, we will first train agents' policies with shared global component to foster coordinated and effective learning. Locally executable policies will be derived subsequently from the trained global policies via policy distillation. Experiments show that our new framework and algorithm can achieve significantly better performance and higher sample efficiency than a cutting-edge baseline on several multi-agent DRL benchmarks.

Successfully navigating a complex environment to obtain a desired outcome is a difficult task, that up to recently was believed to be capable only by humans. This perception has been broken down over time, especially with the introduction of deep reinforcement learning, which has greatly increased the difficulty of tasks that can be automated. However, for traditional reinforcement learning agents this requires an environment to be able to provide frequent extrinsic rewards, which are not known or accessible for many real-world environments. This project aims to explore and contrast existing reinforcement learning solutions that circumnavigate the difficulties of an environment that provide sparse rewards. Different reinforcement solutions will be implemented over a several video game environments with varying difficulty and varying frequency of rewards, as to properly investigate the applicability of these solutions. This project introduces a novel reinforcement learning solution, by combining aspects of two existing state of the art sparse reward solutions.

Recently, neural networks have shown promising results on Document-level Aspect Sentiment Classification (DASC). However, these approaches often offer little transparency w.r.t. their inner working mechanisms and lack interpretability. In this paper, to simulating the steps of analyzing aspect sentiment in a document by human beings, we propose a new Hierarchical Reinforcement Learning (HRL) approach to DASC. This approach incorporates clause selection and word selection strategies to tackle the data noise problem in the task of DASC. First, a high-level policy is proposed to select aspect-relevant clauses and discard noisy clauses. Then, a low-level policy is proposed to select sentiment-relevant words and discard noisy words inside the selected clauses. Finally, a sentiment rating predictor is designed to provide reward signals to guide both clause and word selection. Experimental results demonstrate the impressive effectiveness of the proposed approach to DASC over the state-of-the-art baselines.

Deep Reinforcement Learning (Deep RL) has been receiving increasingly more attention thanks to its encouraging performance on a variety of control tasks. Yet, conventional regularization techniques in training neural networks (e.g., $L_2$ regularization, dropout) have been largely ignored in RL methods, possibly because agents are typically trained and evaluated in the same environment. In this work, we present the first comprehensive study of regularization techniques with multiple policy optimization algorithms on continuous control tasks. Interestingly, we find conventional regularization techniques on the policy networks can often bring large improvement on the task performance, and the improvement is typically more significant when the task is more difficult. We also compare with the widely used entropy regularization and find $L_2$ regularization is generally better. Our findings are further confirmed to be robust against the choice of training hyperparameters. We also study the effects of regularizing different components and find that only regularizing the policy network is typically enough. We hope our study provides guidance for future practices in regularizing policy optimization algorithms.

In this blog post, we explore a functional paradigm for implementing reinforcement learning (RL) algorithms. The paradigm will be that developers write the numerics of their algorithm as independent, pure functions, and then use a library to compile them into policies that can be trained at scale. We share how these ideas were implemented in RLlib's policy builder API, eliminating thousands of lines of "glue" code and bringing support for Keras and TensorFlow 2.0. One of the key ideas behind functional programming is that programs can be composed largely of pure functions, i.e., functions whose outputs are entirely determined by their inputs. Here less is more: by imposing restrictions on what functions can do, we gain the ability to more easily reason about and manipulate their execution.