In this paper, we have provided a new method for the featurization of persistence diagrams through the use of template functions; that is, collections of functions compactly supported on the upper half plane away from the diagonal whose induced fu...

We have introduced a new evaluation method called tre for generating graded judgments about compositional structure in representation learning problems where the structure of the observations is understood. tre infers a set of primitive meaning re...

In this work we propose a spelling correction model which is trained to explicitly correct the errors made by a speech recognizer. To train the SC model, we generate error hypotheses by decoding the TTS data synthesized from a large text-only corp...

Graph Convolutional Networks (GCNs) and their variants have experienced significant attention and have become the de facto methods for learning graph representations. GCNs derive inspiration primarily from recent deep learning approaches, and as a...

We have introduced a new 2v2 soccer domain with simulated physics for continuous multi-agent reinforcement learning research, and used competition between agents in this simple domain to train teams of independent RL agents, demonstrating coordina...

The analysis of the mixture of Gaussians in the preceding section used only standard bounds and simple properties of the normal distribution, so it should be apparent to the reader that many extensions of this result (e.g., Gaussians with a more g...

Our recommendation policy for private types satisfies a relaxed version of the BIC property, called δ-BIC, where the right-hand side in (1) is −δ for some fixed δ>0. We assume a more permissive behavioral model in which agents obey such policy.

The authors of this article thank Prof. Zhiyi Zhang, at UNC Charlotte, for sharing his meaningful thoughts with us during this research.

In this section we compare our proposed algorithm AdaLoss with existing adaptive methods. Algorithm 1 can be viewed as a variant of the standard AdaGrad algorithm proposed by duchi2011adaptive, where the norm version of the update is

In this paper, we proposed a repetition normalized adversarial reward for reinforcement learning on headline generation. We empirically showed that our repetition penalized reward greatly decreased the repetition rate of the generated headlines an...

Metric-based meta-learning techniques have successfully been applied to few-shot classification problems. However, leveraging cross-modal information in a few-shot setting has yet to be explored. When the support from visual information is limited...

In many machine learning applications, especially in healthcare, it is of paramount importance to consider feature acquisition costs. In this paper, we prepared a dataset consisting of nutrition and health data as well as feature acquisition costs...

We proposed a directional regularized tensor modeling for video rain streaks removal in this work. The main contribution is the assumption that all rain streaks fall in an approximately similar arbitrary direction. Experiments show that the propos...

This study presents a novel cross-modal modulating method of vibrotactile fine roughness perception. The user studies yielded the following findings:

Accurate estimation of transmission map is still a challenging problem of common concern in image dehazing. In this work, the coarse transmission map was first generated by weightedly summing up two different transmission maps, respectively, estim...

In this paper, a framework was proposed to mitigate jamming during the uplink transmission of mMIMO systems. The proposed framework exploits the new MMSE-JS estimator and the ZFJS detector during the pilot and data phases, respectively. MMSE-JS us...

We proposed a novel neural network model for simultaneous sentence-level extraction of related pairs. Our model exploits target and context pair-specific representations and creates pair representations that encode up-to l-length walks between the...

In this paper, we propose a Detector-in-Detector network (DID-Net) for the multi-level objects by simply appending a light-weight PartsDetector after Faster RCNN structure to perform parts detection inside each body. We implement our methods on Hu...

Deep Reinforcement Learning has shown great success in a variety of control tasks. However, it is unclear how close we are to the vision of putting Deep RL into practice to solve real world problems. In particular, common practice in the field is ...

We presented an analysis on how to choose the center and axes of compliance for a dual-arm alignment task where the master moves along a linear trajectory and both arms can be set to compliant. In addition, we showed how to easily learn these para...