Abstract-- We propose a hybrid approach aimed at improving thesample efficiency in goal-directed reinforcement learning. We do this via a two-step mechanism where firstly, we approximate a model from Model-Free reinforcement learning. Then, we leverage this approximate model along with a notion of reachability using Mean First Passage Times to perform Model-Based reinforcement learning. Built on such a novel observation, we design two new algorithms - Mean First Passage Time based Q-Learning (MFPT-Q) and Mean First Passage Time based DYNA (MFPT-DYNA), that have been fundamentally modified from the state-of-the-art reinforcement learning techniques. Preliminary results have shown that our hybrid approaches converge with much fewer iterations than their corresponding state-of-the-art counterparts and therefore requiring much fewer samples and much fewer training trials to converge. I. INTRODUCTION Reinforcement Learning (RL) has been successfully applied to numerous challenging problems for autonomous agents to behave intelligently in unstructured real-world environment. One interesting area of research in RL which motivates this work is goal-directed reinforcement learning problem (GDRLP) [1] [2]. In GDRLP, the learning process takes place in two stages.

Real-world tasks are often highly structured. Hierarchical reinforcement learning (HRL) has attracted research interest as an approach for leveraging the hierarchical structure of a given task in reinforcement learning (RL). However, identifying the hierarchical policy structure that enhances the performance of RL is not a trivial task. In this paper, we propose an HRL method that learns a latent variable of a hierarchical policy using mutual information maximization. Our approach can be interpreted as a way to learn a discrete and latent representation of the state-action space. To learn option policies that correspond to modes of the advantage function, we introduce advantage-weighted importance sampling. In our HRL method, the gating policy learns to select option policies based on an option-value function, and these option policies are optimized based on the deterministic policy gradient method. This framework is derived by leveraging the analogy between a monolithic policy in standard RL and a hierarchical policy in HRL by using a deterministic option policy. Experimental results indicate that our HRL approach can learn a diversity of options and that it can enhance the performance of RL in continuous control tasks.

A modification of the confidence screening mechanism based on adaptive weighing of every training instance at each cascade level of the Deep Forest is proposed. The idea underlying the modification is very simple and stems from the confidence screening mechanism idea proposed by Pang et al. to simplify the Deep Forest classifier by means of updating the training set at each level in accordance with the classification accuracy of every training instance. However, if the confidence screening mechanism just removes instances from training and testing processes, then the proposed modification is more flexible and assigns weights by taking into account the classification accuracy. The modification is similar to the AdaBoost to some extent. Numerical experiments illustrate good performance of the proposed modification in comparison with the original Deep Forest proposed by Zhou and Feng.

Abstract--Fundamental to robotics is the debate between model-based and model-free learning: should the robot build an explicit model of the world, or learn a policy directly? In the context of HRI, part of the world to be modeled is the human. One option is for the robot to treat the human as a black box and learn a policy for how they act directly. But it can also model the human as an agent, and rely on a "theory of mind" to guide or bias the learning (grey box). We contribute a characterization of the performance of these methods under the optimistic case of having an ideal theory of mind, as well as under different scenarios in which the assumptions behind the robot's theory of mind for the human are wrong, as they inevitably will be in practice. We find that there is a significant sample complexity advantage to theory of mind methods and that they are more robust to covariate shift, but that when enough interaction data is available, black box approaches eventually dominate. An age-old debate that still animates the halls of computer science, robotics, neuroscience, and psychology departments alike is that between model-based and model-free (reinforcement) learning. Model-based methods work by building a model of the world - the dynamics that tells an agent how the world state will change as a consequence of its actions - and optimizing a cost or reward function under the learned model. In contrast, model-free methods never attempt to explicitly learn how the world works. Instead, the agent learns a policy directly from acting in the world and learning from what works and what does not. Model-free methods are appealing because the agent implicitly learns what it needs to know about the world, and only what it needs. Model-based methods are appealing because knowing how the world works might enable the agent to generalize beyond its experience, and possibly be able to explain why a decision is the best one. In neuro-and cognitive science, the debate is about which paradigm best describes human learning [1], [2].

As sound event classification moves towards larger datasets, issues of label noise become inevitable. Web sites can supply large volumes of user-contributed audio and metadata, but inferring labels from this metadata introduces errors due to unreliable inputs, and limitations in the mapping. There is, however, little research into the impact of these errors. To foster the investigation of label noise in sound event classification we present FSDnoisy18k, a dataset containing 42.5 hours of audio across 20 sound classes, including a small amount of manually-labeled data and a larger quantity of real-world noisy data. We characterize the label noise empirically, and provide a CNN baseline system. Experiments suggest that training with large amounts of noisy data can outperform training with smaller amounts of carefully-labeled data. We also show that noise-robust loss functions can be effective in improving performance in presence of corrupted labels.

Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is that to make the distributed cluster work with high utilization, the workload distributed to each node must be large, which implies nontrivial growth in the SGD mini-batch size. In this paper, we propose a framework called FPDeep, which uses a hybrid of model and layer parallelism to configure distributed reconfigurable clusters to train DNNs. This approach has numerous benefits. First, the design does not suffer from batch size growth. Second, novel workload and weight partitioning leads to balanced loads of both among nodes. And third, the entire system is a fine-grained pipeline. This leads to high parallelism and utilization and also minimizes the time features need to be cached while waiting for back-propagation. As a result, storage demand is reduced to the point where only on-chip memory is used for the convolution layers. We evaluate FPDeep with the Alexnet, VGG-16, and VGG-19 benchmarks. Experimental results show that FPDeep has good scalability to a large number of FPGAs, with the limiting factor being the FPGA-to-FPGA bandwidth. With 6 transceivers per FPGA, FPDeep shows linearity up to 83 FPGAs. Energy efficiency is evaluated with respect to GOPs/J. FPDeep provides, on average, 6.36x higher energy efficiency than comparable GPU servers.

Recently, there has been emerging interest in constructing reproducing kernel Banach spaces (RKBS) for applied and theoretical purposes such as machine learning, sampling reconstruction, sparse approximation and functional analysis. Existing constructions include the reflexive RKBS via a bilinear form, the semi-inner-product RKBS, the RKBS with $\ell^1$ norm, the $p$-norm RKBS via generalized Mercer kernels, etc. The definitions of RKBS and the associated reproducing kernel in those references are dependent on the construction. Moreover, relations among those constructions are unclear. We explore a generic definition of RKBS and the reproducing kernel for RKBS that is independent of construction. Furthermore, we propose a framework of constructing RKBSs that unifies existing constructions mentioned above via a continuous bilinear form and a pair of feature maps. A new class of Orlicz RKBSs is proposed. Finally, we develop representer theorems for machine learning in RKBSs constructed in our framework, which also unifies representer theorems in existing RKBSs.

The model interpretation is essential in many application scenarios and to build a classification model with a ease of model interpretation may provide useful information for further studies and improvement. It is common to encounter with a lengthy set of variables in modern data analysis, especially when data are collected in some automatic ways. This kinds of datasets may not collected with a specific analysis target and usually contains redundant features, which have no contribution to a the current analysis task of interest. Variable selection is a common way to increase the ability of model interpretation and is popularly used with some parametric classification models. There is a lack of studies about variable selection in nonparametric classification models such as the density estimation-based methods and this is especially the case for multiple-class classification situations. In this study we study multiple-class classification problems using the thought of sparse non-parametric density estimation and propose a method for identifying high impacts variables for each class. We present the asymptotic properties and the computation procedure for the proposed method together with some suggested sample size. We also repost the numerical results using both synthesized and some real data sets.

We propose a novel online learning algorithm, called SpCoSLAM 2.0, for spatial concepts and lexical acquisition with high accuracy and scalability. Previously, we proposed SpCoSLAM as an online learning algorithm based on unsupervised Bayesian probabilistic model that integrates multimodal place categorization, lexical acquisition, and SLAM. However, our previous algorithm had limited estimation accuracy owing to the influence of the early stages of learning, and increased computational complexity with added training data. Therefore, we introduce techniques such as fixed-lag rejuvenation to reduce the calculation time while maintaining an accuracy higher than that of the previous algorithm. The results show that, in terms of estimation accuracy, the proposed algorithm exceeds the previous algorithm and is comparable to batch learning. In addition, the calculation time of the proposed algorithm does not depend on the amount of training data and becomes constant for each step of the scalable algorithm. Our approach will contribute to the realization of long-term spatial language interactions between humans and robots.

According to the proposal, the purpose of the restrictions would be to bolster the U.S.'s national security -- after all, AI has many potential military uses, so why would the U.S. want to put that technology in the hands of nations such as China or North Korea? Silicon Valley opposes these AI export restrictions for a few reasons, according to the NYT's story. Perhaps most obviously, they could hurt American companies and help foreign ones. By limiting where companies such as Amazon and Google could sell their AI services, the restrictions would damage those companies' bottom lines. They'll also leave open a market gap that other nations, such as China, could fill.