Recent advances in deep learning have enabled state-of-the-art performance detecting medium and large-size objects. However, small object detection remains challenging primarily due to the scarcity of information. This paper proposes an end-to-end fusion network that integrates hand-crafted features address this limitation. A module based on semantic context information is designed enhance feat...

We have observed that when people engage in algebraic reasoning, they often perceptually and spatially transform algebraic notations directly rather than first converting the notation to an internal, nonspatial representation. We describe empirical evidence for spatial transformations, such as spatially compact grouping, transposition, spatially overlaid intermediate results, cancelling out, sw...

This paper describes a machine learning approach to build an efficient, accurate and fast name spotting system. Finding names in free text is an important task in addressing real-world textbased applications. Most previous approaches have been based on carefully hand-crafted modules encoding linguistic knowledge specific to the language and document genre. Such approaches have two drawbacks: th...

Query expansion is an effective technique to improve the performance of information retrieval systems. Although hand-crafted lexical resources, such as WordNet, could provide more reliable related terms, previous studies showed that query expansion using only WordNet leads to very limited performance improvement. One of the main challenges is how to assign appropriate weights to expanded terms....

We introduce a new large-scale music dataset, MusicNet, to serve as a source of supervision and evaluation of machine learning methods for music research. MusicNet consists of hundreds of freely-licensed classical music recordings by 10 composers, written for 11 instruments, together with instrument/note annotations resulting in over 1 million temporal labels on 34 hours of chamber music perfor...

This paper describes a machine learning approach to build an eÆcient, accurate and fast name spotting system. Finding names in free text is an important task in addressing real-world textbased applications. Most previous approaches have been based on carefully hand-crafted modules encoding linguistic knowledge speci c to the language and document genre. Such approaches have two drawbacks: they ...

In this paper, a hybrid system was proposed for chemical entity mention recognition (CEMP) and gene/protein related object recognition (GPRO) in BeCalm challenge. Firstly, five individual machine learning-based subsystems were developed to identify chemical and gene/protein related entity mentions, that is, a bidirectional LSTM (long-short term memory, a variant of recurrent neural network)-bas...

In recent work, it was shown that combining multi-kernel based support vector machines (SVMs) can lead to near state-of-the-art performance on an action recognition dataset (HMDB-51 dataset). This was 0.4% lower than frameworks that used hand-crafted features in addition to the deep convolutional feature extractors. In the present work, we show that combining distributed Gaussian Processes with...

Several combinatorial optimization problems can be translated into the Weighted Partial Maximum Satisfiability (WPMS) problem. This is an optimization variant of the Satisfiability (SAT) problem. There are two main families of WPMS solvers based on SAT technology: branch and bound and SAT-based. From the MaxSAT evaluations, we have learned that SAT-based solvers dominate on industrial instances...

We have developed a methodology for constructing domain-level expert knowledge bases automatically through crowdsourcing. This approach involves collecting and analyzing the work of numerous students within an intelligent tutor and using an intelligent algorithm to coalesce data to construct the domain model. This evolving expert knowledge base (EEKB) is then utilized to provide expert coaching...