Vision-based grasping systems typically adopt an open-loop execution of a planned grasp. This policy can fail due to many reasons, including ubiquitous calibration error. Recovery from a failed grasp is further complicated by visual occlusion, as the hand is usually occluding the vision sensor as it attempts another open-loop regrasp. This work presents MAT, a tactile closed-loop method capable of realizing grasps provided by a coarse initial positioning of the hand above an object. Our algorithm is a deep reinforcement learning (RL) policy optimized through the clipped surrogate objective within a maximum entropy RL framework to balance exploitation and exploration. The method utilizes tactile and proprioceptive information to act through both fine finger motions and larger regrasp movements to execute stable grasps. A novel curriculum of action motion magnitude makes learning more tractable and helps turn common failure cases into successes. Careful selection of features that exhibit small sim-to-real gaps enables this tactile grasping policy, trained purely in simulation, to transfer well to real world environments without the need for additional learning. Experimentally, this methodology improves over a vision-only grasp success rate substantially on a multi-fingered robot hand. When this methodology is used to realize grasps from coarse initial positions provided by a vision-only planner, the system is made dramatically more robust to calibration errors in the camera-robot transform.

Arguably, intelligent agents ought to be able to discover their own questions so that in learning answers for them they learn unanticipated useful knowledge and skills; this departs from the focus in much of machine learning on agents learning answers to externally defined questions. We present a novel method for a reinforcement learning (RL) agent to discover questions formulated as general value functions or GVFs, a fairly rich form of knowledge representation. Specifically, our method uses non-myopic meta-gradients to learn GVF-questions such that learning answers to them, as an auxiliary task, induces useful representations for the main task faced by the RL agent. We demonstrate that auxiliary tasks based on the discovered GVFs are sufficient, on their own, to build representations that support main task learning, and that they do so better than popular hand-designed auxiliary tasks from the literature. Furthermore, we show, in the context of Atari 2600 videogames, how such auxiliary tasks, meta-learned alongside the main task, can improve the data efficiency of an actor-critic agent.

Mixed cooperative-competitive control scenarios such as human-machine interaction with individual goals of the interacting partners are very challenging for reinforcement learning agents. In order to contribute towards intuitive human-machine collaboration, we focus on problems in the continuous state and control domain where no explicit communication is considered and the agents do not know the others' goals or control laws but only sense their control inputs retrospectively. Our proposed framework combines a learned partner model based on online data with a reinforcement learning agent that is trained in a simulated environment including the partner model. Thus, we overcome drawbacks of independent learners and, in addition, benefit from a reduced amount of real world data required for reinforcement learning which is vital in the human-machine context. We finally analyze an example that demonstrates the merits of our proposed framework which learns fast due to the simulated environment and adapts to the continuously changing partner due to the partner approximation. Keywords: Reinforcement Learning, Mixed Cooperative-Competitive Control, Opponent Modeling.

Probabilistic Boolean Networks (PBNs) were introduced as a computational model for studying gene interactions in Gene Regulatory Networks (GRNs). Controllability of PBNs, and hence GRNs, is the process of making strategic interventions to a network in order to drive it from a particular state towards some other potentially more desirable state. This is of significant importance to systems biology as successful control could be used to obtain potential gene treatments by making therapeutic interventions. Recent advancements in Deep Reinforcement Learning have enabled systems to develop policies merely by interacting with the environment, without complete knowledge of the underlying Markov Decision Process (MDP). In this paper we have implemented a Deep Q Network with Double Q Learning, that directly interacts with the environment -that is, a Probabilistic Boolean Network. Our approach develops a control policy by sampling experiences obtained from the environment using Prioritized Experience Replay which successfully drives a PBN from any state towards the desired one. This novel approach sets the foundations for overcoming the inability to scale to larger PBNs and opens up the spectrum in which to consider control of GRNs without the need of a computational model, i.e. by direct interventions to the GRN.

Deep reinforcement learning (RL) has been a commonly-used strategy for the abstractive summarization task to address both the exposure bias and non-differentiable task issues. However, the conventional reward Rouge-L simply looks for exact n-grams matches between candidates and annotated references, which inevitably makes the generated sentences repetitive and incoherent. In this paper, instead of Rouge-L, we explore the practicability of utilizing the distributional semantics to measure the matching degrees. With distributional semantics, sentence-level evaluation can be obtained, and semantically-correct phrases can also be generated without being limited to the surface form of the reference sentences. Human judgments on Gigaword and CNN/Daily Mail datasets show that our proposed distributional semantics reward (DSR) has distinct superiority in capturing the lexical and compositional diversity of natural language.

This is the next blog in my random series on better understanding some of these advanced Artificial Intelligence and Deep Learning algorithms. This "episode" takes on Generative Adversarial Networks (GANs). Hope you enjoy my "Deep Learning" learning journey. I originally wrote in "Transforming from Autonomous to Smart: Reinforcement Learning Basics" how Reinforcement Learning was creating learning agents to beat games such as Chess, Go and Mario Bros. Reinforcement learning creates intelligent agents that learn via trial-and-error how to map situations to actions so as to maximize rewards. Reinforcement Learning is one of the more powerful Artificial Intelligence (AI) concepts because it is designed to learn and circumnavigate "situations" where you don't have data sets with explicit known outcomes (which represents most real-life situations, like operating an autonomous vehicle).

One of the key issues for imitation learning lies in making policy learned from limited samples to generalize well in the whole state-action space. This problem is much more severe in high-dimensional state environments, such as game playing with raw pixel inputs. Under this situation, even state-of-the-art adversary based imitation learning algorithms fail. Through theoretical and empirical studies, we find that the main cause lies in the failure of training a powerful discriminator to generate meaningful rewards in high-dimensional environments. Theoretical results are provided to suggest the necessity of dimensionality reduction. However, since preserving important discriminative information via feature transformation is a non-trivial task, a straightforward application of off-the-shelf methods cannot achieve desirable performance. To address the above issues, we propose HashReward, which is a novel imitation learning algorithm utilizing the idea of supervised hashing to realize effective training of the discriminator. As far as we are aware, HashReward is the first pure imitation learning approach to achieve expert comparable performance in Atari game environments with raw pixel inputs.

Reinforcement learning algorithms such as the deep deterministic policy gradient algorithm (DDPG) has been widely used in continuous control tasks. However, the model-free DDPG algorithm suffers from high sample complexity. In this paper we consider the deterministic value gradients to improve the sample efficiency of deep reinforcement learning algorithms. Previous works consider deterministic value gradients with the finite horizon, but it is too myopic compared with infinite horizon. We firstly give a theoretical guarantee of the existence of the value gradients in this infinite setting. Based on this theoretical guarantee, we propose a class of the deterministic value gradient algorithm (DVG) with infinite horizon, and different rollout steps of the analytical gradients by the learned model trade off between the variance of the value gradients and the model bias. Furthermore, to better combine the model-based deterministic value gradient estimators with the model-free deterministic policy gradient estimator, we propose the deterministic value-policy gradient (DVPG) algorithm. We finally conduct extensive experiments comparing DVPG with state-of-the-art methods on several standard continuous control benchmarks. Results demonstrate that DVPG substantially outperforms other baselines.

Transfer of Temporal Logic Formulas in Reinforcement Learning Zhe Xu and Ufuk Topcu Abstract Transferring high-level knowledge from a source task to a target task is an effective way to expedite reinforcement learning (RL). For example, propositional logic and first-order logic have been used as representations of such knowledge. We study the transfer of knowledge between tasks in which the timing of the events matters. We call such tasks temporal tasks . We concretize similarity between temporal tasks through a notion of logical transferability, and develop a transfer learning approach between different yet similar temporal tasks. We first propose an inference technique to extract metric interval temporal logic (MITL) formulas in sequential disjunctive normal form from labeled trajectories collected in RL of the two tasks. If logical transferability is identified through this inference, we construct a timed automaton for each sequential conjunctive subformula of the inferred MITL formulas from both tasks. We perform RL on the extended state which includes the locations and clock valuations of the timed automata for the source task. We then establish mappings between the corresponding components (clocks, locations, etc.) of the timed automata from the two tasks, and transfer the extended Q-functions based on the established mappings. Finally, we perform RL on the extended state for the target task, starting with the transferred extended Q-functions. Our results in two case studies show, depending on how similar the source task and the target task are, that the sampling efficiency for the target task can be improved by up to one order of magnitude by performing RL in the extended state space, and further improved by up to another order of magnitude using the transferred extended Q-functions. 1 Introduction Reinforcement learning (RL) has been successful in numerous applications.