The performance of off-policy learning, including deep Q-learning and deep deterministic policy gradient (DDPG), critically depends on the choice of the exploration policy. Existing exploration methods are mostly based on adding noise to the on-going actor policy and can only explore \emph{local} regions close to what the actor policy dictates. In this work, we develop a simple meta-policy gradient algorithm that allows us to adaptively learn the exploration policy in DDPG. Our algorithm allows us to train flexible exploration behaviors that are independent of the actor policy, yielding a \emph{global exploration} that significantly speeds up the learning process. With an extensive study, we show that our method significantly improves the sample-efficiency of DDPG on a variety of reinforcement learning tasks.
Genetic algorithms have been widely used in many practical optimization problems. Inspired by natural selection, operators, including mutation, crossover and selection, provide effective heuristics for search and black-box optimization. However, they have not been shown useful for deep reinforcement learning, possibly due to the catastrophic consequence of parameter crossovers of neural networks. Here, we present Genetic Policy Optimization (GPO), a new genetic algorithm for sample-efficient deep policy optimization. GPO uses imitation learning for policy crossover in the state space and applies policy gradient methods for mutation. Our experiments on MuJoCo tasks show that GPO as a genetic algorithm is able to provide superior performance over the state-of-the-art policy gradient methods and achieves comparable or higher sample efficiency.
Policy gradient methods have achieved remarkable successes in solving challenging reinforcement learning problems. However, it still often suffers from the large variance issue on policy gradient estimation, which leads to poor sample efficiency during training. In this work, we propose a control variate method to effectively reduce variance for policy gradient methods. Motivated by the Stein's identity, our method extends the previous control variate methods used in REINFORCE and advantage actor-critic by introducing more general action-dependent baseline functions. Empirical studies show that our method significantly improves the sample efficiency of the state-of-the-art policy gradient approaches.
Linguistic sequence labeling is a general modeling approach that encompasses a variety of problems, such as part-of-speech tagging and named entity recognition. Recent advances in neural networks (NNs) make it possible to build reliable models without handcrafted features. However, in many cases, it is hard to obtain sufficient annotations to train these models. In this study, we develop a novel neural framework to extract abundant knowledge hidden in raw texts to empower the sequence labeling task. Besides word-level knowledge contained in pre-trained word embeddings, character-aware neural language models are incorporated to extract character-level knowledge. Transfer learning techniques are further adopted to mediate different components and guide the language model towards the key knowledge. Comparing to previous methods, these task-specific knowledge allows us to adopt a more concise model and conduct more efficient training. Different from most transfer learning methods, the proposed framework does not rely on any additional supervision. It extracts knowledge from self-contained order information of training sequences. Extensive experiments on benchmark datasets demonstrate the effectiveness of leveraging character-level knowledge and the efficiency of co-training. For example, on the CoNLL03 NER task, model training completes in about 6 hours on a single GPU, reaching F1 score of 91.71$\pm$0.10 without using any extra annotation.
We propose a new localized inference algorithm for answering marginalization queries in large graphical models with the correlation decay property. Given a query variable and a large graphical model, we define a much smaller model in a local region around the query variable in the target model so that the marginal distribution of the query variable can be accurately approximated. We introduce two approximation error bounds based on the Dobrushin's comparison theorem and apply our bounds to derive a greedy expansion algorithm that efficiently guides the selection of neighbor nodes for localized inference. We verify our theoretical bounds on various datasets and demonstrate that our localized inference algorithm can provide fast and accurate approximation for large graphical models.
Deep neural networks have been remarkable successful in various AI tasks but often cast high computation and energy cost for energy-constrained applications such as mobile sensing. We address this problem by proposing a novel framework that optimizes the prediction accuracy and energy cost simultaneously, thus enabling effective cost-accuracy trade-off at test time. In our framework, each data instance is pushed into a cascade of deep neural networks with increasing sizes, and a selection module is used to sequentially determine when a sufficiently accurate classifier can be used for this data instance. The cascade of neural networks and the selection module are jointly trained in an end-to-end fashion by the REINFORCE algorithm to optimize a trade-off between the computational cost and the predictive accuracy. Our method is able to simultaneously improve the accuracy and efficiency by learning to assign easy instances to fast yet sufficiently accurate classifiers to save computation and energy cost, while assigning harder instances to deeper and more powerful classifiers to ensure satisfiable accuracy. With extensive experiments on several image classification datasets using cascaded ResNet classifiers, we demonstrate that our method outperforms the standard well-trained ResNets in accuracy but only requires less than 20% and 50% FLOPs cost on the CIFAR-10/100 datasets and 66% on the ImageNet dataset, respectively.
Human vision possesses strong invariance in image recognition. The cognitive capability of deep convolutional neural network (DCNN) is close to the human visual level because of hierarchical coding directly from raw image. Owing to its superiority in feature representation, DCNN has exhibited remarkable performance in scene recognition of high-resolution remote sensing (HRRS) images and classification of hyper-spectral remote sensing images. In-depth investigation is still essential for understanding why DCNN can accurately identify diverse ground objects via its effective feature representation. Thus, we train the deep neural network called AlexNet on our large scale remote sensing image recognition benchmark. At the neuron level in each convolution layer, we analyze the general properties of DCNN in HRRS image recognition by use of a framework of visual stimulation-characteristic response combined with feature coding-classification decoding. Specifically, we use histogram statistics, representational dissimilarity matrix, and class activation mapping to observe the selective and invariance representations of DCNN in HRRS image recognition. We argue that selective and invariance representations play important roles in remote sensing images tasks, such as classification, detection, and segment. Also selective and invariance representations are significant to design new DCNN liked models for analyzing and understanding remote sensing images.
Recently, deep convolutional neural network (DCNN) achieved increasingly remarkable success and rapidly developed in the field of natural image recognition. Compared with the natural image, the scale of remote sensing image is larger and the scene and the object it represents are more macroscopic. This study inquires whether remote sensing scene and natural scene recognitions differ and raises the following questions: What are the key factors in remote sensing scene recognition? Is the DCNN recognition mechanism centered on object recognition still applicable to the scenarios of remote sensing scene understanding? We performed several experiments to explore the influence of the DCNN structure and the scale of remote sensing scene understanding from the perspective of scene complexity. Our experiment shows that understanding a complex scene depends on an in-depth network and multiple-scale perception. Using a visualization method, we qualitatively and quantitatively analyze the recognition mechanism in a complex remote sensing scene and demonstrate the importance of multi-objective joint semantic support.
Policy gradient methods have been successfully applied to many complex reinforcement learning problems. However, policy gradient methods suffer from high variance, slow convergence, and inefficient exploration. In this work, we introduce a maximum entropy policy optimization framework which explicitly encourages parameter exploration, and show that this framework can be reduced to a Bayesian inference problem. We then propose a novel Stein variational policy gradient method (SVPG) which combines existing policy gradient methods and a repulsive functional to generate a set of diverse but well-behaved policies. SVPG is robust to initialization and can easily be implemented in a parallel manner. On continuous control problems, we find that implementing SVPG on top of REINFORCE and advantage actor-critic algorithms improves both average return and data efficiency.
We study the interpretability of conditional probability estimates for binary classification under the agnostic setting or scenario. Under the agnostic setting, conditional probability estimates do not necessarily reflect the true conditional probabilities. Instead, they have a certain calibration property: among all data points that the classifier has predicted P(Y = 1|X) = p, p portion of them actually have label Y = 1. For cost-sensitive decision problems, this calibration property provides adequate support for us to use Bayes Decision Theory. In this paper, we define a novel measure for the calibration property together with its empirical counterpart, and prove an uniform convergence result between them. This new measure enables us to formally justify the calibration property of conditional probability estimations, and provides new insights on the problem of estimating and calibrating conditional probabilities.