In this paper, we proposed a scalable approach to increase diversity in mini-batch Active Learning, and linked the approach to Facility Location. We have experimented with the datasets of various sizes and nature, using models of various complexit...

In this paper we have presented a novel approach to brand-influencer matchmaking based on well-known machine learning techniques and using an image-based analysis of user profile content. The results indicated that our algorithm, when presented wi...

We introduced a novel policy for the stochastic convex online learning problem based on constructing a search tree and inducing an efficient random walk on the tree. We established a finite-time regret analysis of the proposed policy which ensures...

In this paper, we have presented a framework to improve semantic segmentation performance at nighttime with stylized data and unlabeled real data of increasing darkness, as well as a novel uncertainty-aware evaluation for semantic segmentation des...

In this paper, we propose the foreground-aware image inpainting model for challenging scenarios involving prediction of both foreground and background pixels. Our model first infers the contours of the objects in the image, then uses the completed...

Given the worldwide need for user accessibility and UD in assistive robotics, this work provided a pioneer study of end-user interest, comprehension and satisfaction regarding the reproduction of sign language by a humanoid robot.

Inter-robot loop closure detection is a critical component for many multirobot applications, including those that require collaborative localization and mapping. The resource-intensive nature of this process, together with the limited nature of mi...

In this paper, based on the general linear regression model, we proposed a new manifold regularized multi-task learning model for a joint analysis of multiple modalities. Instead of including all high-dimensional features to predict performance, o...

We have proposed a particular neural network architecture aimed at spatiotemporal modeling with cyclical components applied to the example of estimating demand, which is an important problem in the ridesharing space. We furthermore perform Bayesia...

We have developed a new method for the determination of plane directions in APT datasets. Here we have shown that this can enhance the detection of crystallographic information, above and beyond that of existing algorithms. Simultaneously we have ...

In this paper, we propose an elastic architecture transfer mechanism for accelerating large-scale neural architecture search (EAT-NAS). Rather than spending a lot of computational resources to directly search the neural architecture on large-scale...

In this paper, we have proposed a framework for high-density compression of ultrasound volumes. This framework involves two parts, a compressor with a segmentation block and a decompressor with a generation block. Both the segmentation and generat...

In this paper, we proposed AIE by combining SIL and RND. In addition, we proposed AIE2 and AIE3, which can lead to efficient deep exploration. AIE2 gives an intrinsic penalty reward to states where the agent frequently visits, which prevents the a...

In this work, we used a simple Bi-LSTM model together with embedding from language models (ELMo), achieving state-of-the-art performance on many Chinese word segmentation datasets. Also, we found that our model is more robust when dealing with the...

Semantic segmentation is a critical task to enable fully automated driving. It is also a complex task and requires large amounts of annotated data which is expensive. Large annotated datasets is currently the bottleneck for achieving high accuracy...

In this paper, we presented a method to reconstruct dense large-scale semantic maps based on a monocular camera. The semantic and depth are first predicted jointly by a novel multi-task network. The proposed network improves results with ASPP modu...

In this paper, we propose an ensemble network (EnsembleNet) and explored its priority. The designed EnsembleNet is based on ResNet-50. The features extracted from the multiple branches are concatenated as the final representation of a pedestrian. ...

We developed a synthetic approach based on CycleGAN to produce SynPlanCT images from CBCT images. The proposed approach relies only on unpaired and unaligned CBCT and PlanCT images for training. The image quality of the synthesized PlanCT images s...

In this paper, we present a novel scale aware feature encoder (SAFE) to tackle the problem of having characters with different scales in the text images. SAFE is composed of a multi-scale convolutional encoder and a scale attention network. It can...

We have proposed ANPs, which augment NPs with attention to resolve the fundamental problem of underfitting. We have shown that this greatly improves the accuracy of predictions in terms of context and target NLL, results in faster training, and ex...