In this paper, we utilize deep learning to learn high-level features for audio chord detection. The learned features, obtained by a deep network in bottleneck architecture, give promising results and outperform state-of-the-art systems. We present and evaluate the results for various methods and configurations, including input pre-processing, a bottleneck architecture, and SVMs vs. HMMs for cho...

In this paper, we present a novel approach to automatic 3D Facial Expression Recognition (FER) based on deep representation of facial 3D geometric and 2D photometric attributes. A 3D face is firstly represented by its geometric and photometric attributes, including the geometry map, normal maps, normalized curvature map and texture map. These maps are then fed into a pre-trained deep convolutio...

Theoretical results suggest that in order to learn the kind of complicated functions that can represent highlevel abstractions (e.g. in vision, language, and other AI-level tasks), one may need deep architectures. Deep architectures are composed of multiple levels of non-linear operations, such as in neural nets with many hidden layers or in complicated propositional formulae re-using many sub-...

In past few years, deep learning has received attention in the field of artificial intelligence. This paper reviews three focus areas of learning methods in deep learning namely supervised, unsupervised and reinforcement learning. These learning methods are used in implementing deep and convolutional neural networks. They offered unified computational approach, flexibility and scalability capab...

“Deep-syntactic” dependency structures bridge the gap between the surface-syntactic structures as produced by state-of-the-art dependency parsers and semantic logical forms in that they abstract away from surfacesyntactic idiosyncrasies, but still keep the linguistic structure of a sentence. They have thus a great potential for such downstream applications as machine translation and summarizati...

In recent years, new neural network models with deep architectures started to get more attention in the field of machine learning. These models contain larger number of layers (therefore ”deep”) than conventional multi-layered perceptron, which usually uses only two or three functional layers of neurons. To overcome the difficulties of training such complex networks, new learning algorithms hav...

We have proposed Hidden Conditional Neural Fields (HCNF) for automatic speech recognition and shown the effectiveness by continuous phoneme recognition experiments on the TIMIT and the Japanese ASJ+JNAS corpora. In this paper, we propose to use an observation function with a deep structure in HCNF. The proposed deep observation function enables to use the deep neural networks in HCNF, which hav...

In this paper, we present a framework of a factored deep convolutional neural network (CNN) learning for noise robust automatic speech recognition (ASR). Deep CNN architecture, which has attracted great attention in various research areas, has also been successfully applied to ASR. However, to ensure noise robustness, since merely introducing deep CNN architecture into the acoustic modeling of ...