Solving sparse reward tasks through exploration is one of the major challenges in deep reinforcement learning, especially in three-dimensional, partially-observable environments. Critically, the algorithm proposed in this article uses a single human demonstration to solve hard-exploration problems. We train an agent on a combination of demonstrations and own experience to solve problems with variable initial conditions. We adapt this idea and integrate it with the proximal policy optimization (PPO). The agent is able to increase its performance and to tackle harder problems by replaying its own past trajectories prioritizing them based on the obtained reward and the maximum value of the trajectory. We compare different variations of this algorithm to behavioral cloning on a set of hard-exploration tasks in the Animal-AI Olympics environment. To the best of our knowledge, learning a task in a three-dimensional environment with comparable difficulty has never been considered before using only one human demonstration.
相關內容
Traffic simulators act as an essential component in the operating and planning of transportation systems. Conventional traffic simulators usually employ a calibrated physical car-following model to describe vehicles' behaviors and their interactions with traffic environment. However, there is no universal physical model that can accurately predict the pattern of vehicle's behaviors in different situations. A fixed physical model tends to be less effective in a complicated environment given the non-stationary nature of traffic dynamics. In this paper, we formulate traffic simulation as an inverse reinforcement learning problem, and propose a parameter sharing adversarial inverse reinforcement learning model for dynamics-robust simulation learning. Our proposed model is able to imitate a vehicle's trajectories in the real world while simultaneously recovering the reward function that reveals the vehicle's true objective which is invariant to different dynamics. Extensive experiments on synthetic and real-world datasets show the superior performance of our approach compared to state-of-the-art methods and its robustness to variant dynamics of traffic.
Block-based programming languages like Scratch have become increasingly popular as introductory languages for novices. These languages are intended to be used with a "tinkering" approach which allows learners and teachers to quickly assemble working programs and games, but this often leads to low code quality. Such code can be hard to comprehend, changing it is error-prone, and learners may struggle and lose interest. The general solution to improve code quality is to refactor the code. However, Scratch lacks many of the common abstraction mechanisms used when refactoring programs written in higher programming languages. In order to improve Scratch code, we therefore propose a set of atomic code transformations to optimise readability by (1) rewriting control structures and (2) simplifying scripts using the inherently concurrent nature of Scratch programs. By automating these transformations it is possible to explore the space of possible variations of Scratch programs. In this paper, we describe a multi-objective search-based approach that determines sequences of code transformations which improve the readability of a given Scratch program and therefore form refactorings. Evaluation on a random sample of 1000 Scratch programs demonstrates that the generated refactorings reduce complexity and entropy in 70.4% of the cases, and 354 projects are improved in at least one metric without making any other metric worse. The refactored programs can help both novices and their teachers to improve their code.
While deep reinforcement learning has achieved promising results in challenging decision-making tasks, the main bones of its success --- deep neural networks are mostly black-boxes. A feasible way to gain insight into a black-box model is to distill it into an interpretable model such as a decision tree, which consists of if-then rules and is easy to grasp and be verified. However, the traditional model distillation is usually a supervised learning task under a stationary data distribution assumption, which is violated in reinforcement learning. Therefore, a typical policy distillation that clones model behaviors with even a small error could bring a data distribution shift, resulting in an unsatisfied distilled policy model with low fidelity or low performance. In this paper, we propose to address this issue by changing the distillation objective from behavior cloning to maximizing an advantage evaluation. The novel distillation objective maximizes an approximated cumulative reward and focuses more on disastrous behaviors in critical states, which controls the data shift effect. We evaluate our method on several Gym tasks, a commercial fight game, and a self-driving car simulator. The empirical results show that the proposed method can preserve a higher cumulative reward than behavior cloning and learn a more consistent policy to the original one. Moreover, by examining the extracted rules from the distilled decision trees, we demonstrate that the proposed method delivers reasonable and robust decisions.
The difficulty in specifying rewards for many real-world problems has led to an increased focus on learning rewards from human feedback, such as demonstrations. However, there are often many different reward functions that explain the human feedback, leaving agents with uncertainty over what the true reward function is. While most policy optimization approaches handle this uncertainty by optimizing for expected performance, many applications demand risk-averse behavior. We derive a novel policy gradient-style robust optimization approach, PG-BROIL, that optimizes a soft-robust objective that balances expected performance and risk. To the best of our knowledge, PG-BROIL is the first policy optimization algorithm robust to a distribution of reward hypotheses which can scale to continuous MDPs. Results suggest that PG-BROIL can produce a family of behaviors ranging from risk-neutral to risk-averse and outperforms state-of-the-art imitation learning algorithms when learning from ambiguous demonstrations by hedging against uncertainty, rather than seeking to uniquely identify the demonstrator's reward function.
Interpretation of Deep Neural Networks (DNNs) training as an optimal control problem with nonlinear dynamical systems has received considerable attention recently, yet the algorithmic development remains relatively limited. In this work, we make an attempt along this line by reformulating the training procedure from the trajectory optimization perspective. We first show that most widely-used algorithms for training DNNs can be linked to the Differential Dynamic Programming (DDP), a celebrated second-order trajectory optimization algorithm rooted in the Approximate Dynamic Programming. In this vein, we propose a new variant of DDP that can accept batch optimization for training feedforward networks, while integrating naturally with the recent progress in curvature approximation. The resulting algorithm features layer-wise feedback policies which improve convergence rate and reduce sensitivity to hyper-parameter over existing methods. We show that the algorithm is competitive against state-ofthe-art first and second order methods. Our work opens up new avenues for principled algorithmic design built upon the optimal control theory.
Proximal Policy Optimization (PPO) is a highly popular model-free reinforcement learning (RL) approach. However, in continuous state and actions spaces and a Gaussian policy -- common in computer animation and robotics -- PPO is prone to getting stuck in local optima. In this paper, we observe a tendency of PPO to prematurely shrink the exploration variance, which naturally leads to slow progress. Motivated by this, we borrow ideas from CMA-ES, a black-box optimization method designed for intelligent adaptive Gaussian exploration, to derive PPO-CMA, a novel proximal policy optimization approach that can expand the exploration variance on objective function slopes and shrink the variance when close to the optimum. This is implemented by using separate neural networks for policy mean and variance and training the mean and variance in separate passes. Our experiments demonstrate a clear improvement over vanilla PPO in many difficult OpenAI Gym MuJoCo tasks.
Machine learning algorithms have found several applications in the field of robotics and control systems. The control systems community has started to show interest towards several machine learning algorithms from the sub-domains such as supervised learning, imitation learning and reinforcement learning to achieve autonomous control and intelligent decision making. Amongst many complex control problems, stable bipedal walking has been the most challenging problem. In this paper, we present an architecture to design and simulate a planar bipedal walking robot(BWR) using a realistic robotics simulator, Gazebo. The robot demonstrates successful walking behaviour by learning through several of its trial and errors, without any prior knowledge of itself or the world dynamics. The autonomous walking of the BWR is achieved using reinforcement learning algorithm called Deep Deterministic Policy Gradient(DDPG). DDPG is one of the algorithms for learning controls in continuous action spaces. After training the model in simulation, it was observed that, with a proper shaped reward function, the robot achieved faster walking or even rendered a running gait with an average speed of 0.83 m/s. The gait pattern of the bipedal walker was compared with the actual human walking pattern. The results show that the bipedal walking pattern had similar characteristics to that of a human walking pattern.
Tracking by detection is a common approach to solving the Multiple Object Tracking problem. In this paper we show how deep metric learning can be used to improve three aspects of tracking by detection. We train a convolutional neural network to learn an embedding function in a Siamese configuration on a large person re-identification dataset offline. It is then used to improve the online performance of tracking while retaining a high frame rate. We use this learned appearance metric to robustly build estimates of pedestrian's trajectories in the MOT16 dataset. In breaking with the tracking by detection model, we use our appearance metric to propose detections using the predicted state of a tracklet as a prior in the case where the detector fails. This method achieves competitive results in evaluation, especially among online, real-time approaches. We present an ablative study showing the impact of each of the three uses of our deep appearance metric.
We present an end-to-end framework for solving the Vehicle Routing Problem (VRP) using reinforcement learning. In this approach, we train a single model that finds near-optimal solutions for problem instances sampled from a given distribution, only by observing the reward signals and following feasibility rules. Our model represents a parameterized stochastic policy, and by applying a policy gradient algorithm to optimize its parameters, the trained model produces the solution as a sequence of consecutive actions in real time, without the need to re-train for every new problem instance. On capacitated VRP, our approach outperforms classical heuristics and Google's OR-Tools on medium-sized instances in solution quality with comparable computation time (after training). We demonstrate how our approach can handle problems with split delivery and explore the effect of such deliveries on the solution quality. Our proposed framework can be applied to other variants of the VRP such as the stochastic VRP, and has the potential to be applied more generally to combinatorial optimization problems.
Mobile network that millions of people use every day is one of the most complex systems in real world. Optimization of mobile network to meet exploding customer demand and reduce CAPEX/OPEX poses greater challenges than in prior works. Actually, learning to solve complex problems in real world to benefit everyone and make the world better has long been ultimate goal of AI. However, application of deep reinforcement learning (DRL) to complex problems in real world still remains unsolved, due to imperfect information, data scarcity and complex rules in real world, potential negative impact to real world, etc. To bridge this reality gap, we propose a sim-to-real framework to direct transfer learning from simulation to real world without any training in real world. First, we distill temporal-spatial relationships between cells and mobile users to scalable 3D image-like tensor to best characterize partially observed mobile network. Second, inspired by AlphaGo, we introduce a novel self-play mechanism to empower DRL agents to gradually improve intelligence by competing for best record on multiple tasks, just like athletes compete for world record in decathlon. Third, a decentralized DRL method is proposed to coordinate multi-agents to compete and cooperate as a team to maximize global reward and minimize potential negative impact. Using 7693 unseen test tasks over 160 unseen mobile networks in another simulator as well as 6 field trials on 4 commercial mobile networks in real world, we demonstrate the capability of this sim-to-real framework to direct transfer the learning not only from one simulator to another simulator, but also from simulation to real world. This is the first time that a DRL agent successfully transfers its learning directly from simulation to very complex real world problems with imperfect information, complex rules, huge state/action space, and multi-agent interactions.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191