Deep Reinforcement Learning (DRL) has shown its promising capabilities to learn optimal policies directly from trial and error. However, learning can be hindered if the goal of the learning, defined by the reward function, is "not optimal". We demonstrate that by setting the goal/target of competition in a counter-intuitive but intelligent way, instead of heuristically trying solutions through many hours the DRL simulation can quickly converge into a winning strategy. The ICRA-DJI RoboMaster AI Challenge is a game of cooperation and competition between robots in a partially observable environment, quite similar to the Counter-Strike game. Unlike the traditional approach to games, where the reward is given at winning the match or hitting the enemy, our DRL algorithm rewards our robots when in a geometric-strategic advantage, which implicitly increases the winning chances. Furthermore, we use Deep Q Learning (DQL) to generate multi-agent paths for moving, which improves the cooperation between two robots by avoiding the collision. Finally, we implement a variant A* algorithm with the same implicit geometric goal as DQL and compare results. We conclude that a well-set goal can put in question the need for learning algorithms, with geometric-based searches outperforming DQL in many orders of magnitude.

Combining No-regret and Q-learning Ian A. Kash University of Illinois, Chicago, IL Michael Sullins University of Illinois, Chicago, IL Katja Hofmann Microsoft Research, Cambridge, UK Abstract Counterfactual Regret Minimization (CFR) has found success in settings like poker which have both terminal states and perfect recall. We seek to understand how to relax these requirements. As a first step, we introduce a simple algorithm, local no-regret learning (LONR), which uses a Q-learning-like update rule to allow learning without terminal states or perfect recall. We prove its convergence for the basic case of MDPs (and limited extensions of them) and present empirical results showing that it achieves last iterate convergence in a number of settings, most notably NoSDE games, a class of Markov games specifically designed to be challenging to learn where no prior algorithm is known to achieve convergence to a stationary equilibrium even on average. 1 Introduction V ersions of counterfactual regret minimization (CFR) [50] have found success in playing poker at human expert level [10, 41] as well as fully solving nontrivial versions of it [8]. CFR more generally can solve extensive form games of incomplete information. It works by using a no-regret algorithm to select actions. In particular, one copy of such an algorithm is used at each information set, which corresponds to the full history of play observed by a single agent. The resulting algorithm satisfies a global no-regret guarantee, so at least in two-player zero-sum games is guaranteed to converge to an optimal strategy through sufficient self-play. However, CFR does have limitations. It makes two strong assumptions which are natural for games such as poker, but limit applicability to further settings. First, it assumes that the agent has perfect recall, which in a more general context means that the state representation captures the full history of states visited (and so imposes a tree structure). Current RL domains may rarely repeat states due to their large state spaces, but they certainly do not encode the full history of states and actions. Second, it assumes that a terminal state is eventually reached and performs updates only after this occurs.

Causal learning is a beneficial approach to analyze the cause and effect relationships among variables in a dataset. A causal graph can be generated from a dataset using a particular causal algorithm, for instance, the PC algorithm or Fast Causal Inference (FCI). Generating a causal graph from a dataset that contains different data types (mixed data) is not trivial. This research offers an easy way to handle the mixed data so that it can be used to learn causal graphs using the existing application of the PC algorithm and FCI. This research proposes using kernel functions and Kernel Alignment to handle a mixed data. Two main steps of this approach are computing a kernel matrix for each variable and calculating a pseudo-correlation matrix using Kernel Alignment. Kernel Alignment is used as a substitute for the correlation matrix for the conditional independence test for Gaussian data in PC Algorithm and FCI. The advantage of this idea is that is possible to handle any data type by using a suitable kernel function to compute a kernel matrix for an observed variable. The proposed method is successfully applied to learn a causal graph from a mixed data containing categorical, binary, ordinal, and continuous variables.

Gunrock is the winner of the 2018 Amazon Alexa Prize, as evaluated by coherence and engagement from both real users and Amazon-selected expert conversationalists. We focus on understanding complex sentences and having in-depth conversations in open domains. In this paper, we introduce some innovative system designs and related validation analysis. Overall, we found that users produce longer sentences to Gunrock, which are directly related to users' engagement (e.g., ratings, number of turns). Additionally, users' backstory queries about Gunrock are positively correlated to user satisfaction. Finally, we found dialog flows that interleave facts and personal opinions and stories lead to better user satisfaction.

Over the past decade, the NASA Autonomous Systems and Operations (ASO) project has developed and demonstrated numerous autonomy enabling technologies employing AI techniques. Our work has employed AI in three distinct ways to enable autonomous mission operations capabilities. Crew Autonomy gives astronauts tools to assist in the performance of each of these mission operations functions. Vehicle System Management uses AI techniques to turn the astronaut's spacecraft into a robot, allowing it to operate when astronauts are not present, or to reduce astronaut workload. AI technology also enables Autonomous Robots as crew assistants or proxies when the crew are not present. We first describe human spaceflight mission operations capabilities. We then describe the ASO project, and the development and demonstration performed by ASO since 2011. We will describe the AI techniques behind each of these demonstrations, which include a variety of symbolic automated reasoning and machine learning based approaches. Finally, we conclude with an assessment of future development needs for AI to enable NASA's future Exploration missions.

Automatic KB completion for commonsense knowledge graphs (e.g., ATOMIC and ConceptNet) poses unique challenges compared to the much studied conventional knowledge bases (e.g., Freebase). Commonsense knowledge graphs use free-form text to represent nodes, resulting in orders of magnitude more nodes compared to conventional KBs (18x more nodes in ATOMIC compared to Freebase (FB15K-237)). Importantly, this implies significantly sparser graph structures - a major challenge for existing KB completion methods that assume densely connected graphs over a relatively smaller set of nodes. In this paper, we present novel KB completion models that can address these challenges by exploiting the structural and semantic context of nodes. Specifically, we investigate two key ideas: (1) learning from local graph structure, using graph convolutional networks and automatic graph densification and (2) transfer learning from pre-trained language models to knowledge graphs for enhanced contextual representation of knowledge. We describe our method to incorporate information from both these sources in a joint model and provide the first empirical results for KB completion on ATOMIC and evaluation with ranking metrics on ConceptNet. Our results demonstrate the effectiveness of language model representations in boosting link prediction performance and the advantages of learning from local graph structure (+1.5 points in MRR for ConceptNet) when training on subgraphs for computational efficiency. Further analysis on model predictions shines light on the types of commonsense knowledge that language models capture well.

Failures are challenging for learning to control physical systems since they risk damage, time-consuming resets, and often provide little gradient information. Adding safety constraints to exploration typically requires a lot of prior knowledge and domain expertise. We present a safety measure which implicitly captures how the system dynamics relate to a set of failure states. Not only can this measure be used as a safety function, but also to directly compute the set of safe state-action pairs. Further, we show a model-free approach to learn this measure by active sampling using Gaussian processes. While safety can only be guaranteed after learning the safety measure, we show that failures can already be greatly reduced by using the estimated measure during learning.

Abstract: We propose an architecture for learning complex controllable behaviors by having simple Policies Modulate Trajectory Generators (PMTG), a powerful combination that can provide both memory and prior knowledge to the controller. The result is a flexible architecture that is applicable to a class of problems with periodic motion for which one has an insight into the class of trajectories that might lead to a desired behavior. We illustrate the basics of our architecture using a synthetic control problem, then go on to learn speed-controlled locomotion for a quadrupedal robot by using Deep Reinforcement Learning and Evolutionary Strategies. We demonstrate that a simple linear policy, when paired with a parametric Trajectory Generator for quadrupedal gaits, can induce walking behaviors with controllable speed from 4 -dimensional IMU observations alone, and can be learned in under 1000 rollouts. We also transfer these policies to a real robot and show locomotion with controllable forward velocity. Keywords: Reinforcement Learning, Control, Locomotion 1 Introduction The recent success of Deep Learning (DL) on simulated robotic tasks has opened an exciting research direction. Nevertheless, many robotic tasks such as locomotion still remain an open problem for learning-based methods due to their complexity or dynamics. From a Deep Learning (DL) perspective, one way to tackle these complex problems is by using more and more complex policies (such as recurrent networks). Unfortunately, more complex policies are harder to train and require even more training data which is often problematic for robotics.

Multi-hop question answering requires models to gather information from different parts of a text to answer a question. Most current approaches learn to address this task in an end-to-end way with neural networks, without maintaining an explicit representation of the reasoning process. We propose a method to extract a discrete reasoning chain over the text, which consists of a series of sentences leading to the answer. We then feed the extracted chains to a BERT -based QA model (Devlin et al., 2018) to do final answer prediction. Critically, we do not rely on gold annotated chains or "supporting facts:" at training time, we derive pseudogold reasoning chains using heuristics based on named entity recognition and coreference resolution. Nor do we rely on these annotations at test time, as our model learns to extract chains from raw text alone. We test our approach on two recently proposed large multi-hop question answering datasets: WikiHop (Welbl et al., 2018) and HotpotQA (Y ang et al., 2018), and achieve state-of-art performance on WikiHop and strong performance on HotpotQA. Our analysis shows properties of chains that are crucial for high performance: in particular, modeling extraction sequentially is important, as is dealing with each candidate sentence in a context-aware way. Furthermore, human evaluation shows that our extracted chains allow humans to give answers with high confidence, indicating that these are a strong intermediate abstraction for this task. 1 Introduction As high performance has been achieved in simple question answering settings (Rajpurkar et al., 2016), work on question answering has increasingly gravitated towards questions that require more complex reasoning to solve. Multi-hop question answering datasets explicitly require aggregating clues from different parts of some given documents (Dua et al., 2019; Welbl et al., 2018; Y ang et al., 2018; Jansen et al., 2018; Khashabi et al., 2018).

We aim to guard swarm-robotics applications against denial-of-service (DoS) failures/attacks that result in withdrawals of robots. We focus on applications requiring the selection of actions for each robot, among a set of available ones, e.g., which trajectory to follow. Such applications are central in large-scale robotic/control applications, e.g., multi-robot motion planning for target tracking. But the current attack-robust algorithms are centralized, and scale quadratically with the problem size (e.g., number of robots). Thus, in this paper, we propose a general-purpose distributed algorithm towards robust optimization at scale, with local communications only. We name it distributed robust maximization (DRM). DRM proposes a divide-and-conquer approach that distributively partitions the problem among K cliques of robots. The cliques optimize in parallel, independently of each other. That way, DRM also offers significant computational speed-ups up to 1/K^2 the running time of its centralized counterparts. K depends on the robots' communication range, which is given as input to DRM. DRM also achieves a close-to-optimal performance, equal to the guaranteed performance of its centralized counterparts. We demonstrate DRM's performance in both Gazebo and MATLAB simulations, in scenarios of active target tracking with swarms of robots. We observe DRM achieves significant computational speed-ups (it is 3 to 4 orders faster) and, yet, nearly matches the tracking performance of its centralized counterparts.