arXiv subjects

Li Deng

Publications and source records attributed to Li Deng.

85 records · Page 5Linked to original sources

arXiv2015

End-to-end Learning of LDA by Mirror-Descent Back Propagation over a Deep Architecture

We develop a fully discriminative learning approach for supervised Latent Dirichlet Allocation (LDA) model using Back Propagation (i.e., BP-sLDA), which maximizes the posterior probability of the prediction variable given the input document. Different from traditional variational learning or Gibbs sampling approaches, the proposed learning method applies (i) the mirror descent algorithm for maximum a posterior inference and (ii) back propagation over a deep architecture together with stochastic gradient/mirror descent for model parameter estimation, leading to scalable and end-to-end discriminative learning of the model. As a byproduct, we also apply this technique to develop a new learning method for the traditional unsupervised LDA model (i.e., BP-LDA). Experimental results on three real-world regression and classification tasks show that the proposed methods significantly outperform the previous supervised topic models, neural networks, and is on par with deep neural networks.

cs.LG↗

arXiv2015

Language Models for Image Captioning: The Quirks and What Works

Two recent approaches have achieved state-of-the-art results in image captioning. The first uses a pipelined process where a set of candidate words is generated by a convolutional neural network (CNN) trained on images, and then a maximum entropy (ME) language model is used to arrange these words into a coherent sentence. The second uses the penultimate activation layer of the CNN as input to a recurrent neural network (RNN) that then generates the caption sequence. In this paper, we compare the merits of these different language modeling approaches for the first time by using the same state-of-the-art CNN as input. We examine issues in the different approaches, including linguistic irregularities, caption repetition, and data set overlap. By combining key aspects of the ME and RNN methods, we achieve a new record performance over previously published results on the benchmark COCO dataset. However, the gains we see in BLEU do not translate to human judgments.

cs.CL↗

arXiv2015

Embedding Entities and Relations for Learning and Inference in Knowledge Bases

We consider learning representations of entities and relations in KBs using the neural-embedding approach. We show that most existing models, including NTN (Socher et al., 2013) and TransE (Bordes et al., 2013b), can be generalized under a unified learning framework, where entities are low-dimensional vectors learned from a neural network and relations are bilinear and/or linear mapping functions. Under this framework, we compare a variety of embedding models on the link prediction task. We show that a simple bilinear formulation achieves new state-of-the-art results for the task (achieving a top-10 accuracy of 73.2% vs. 54.7% by TransE on Freebase). Furthermore, we introduce a novel approach that utilizes the learned relation embeddings to mine logical rules such as "BornInCity(a,b) and CityInCountry(b,c) => Nationality(a,c)". We find that embeddings learned from the bilinear objective are particularly good at capturing relational semantics and that the composition of relations is characterized by matrix multiplication. More interestingly, we demonstrate that our embedding-based rule extraction approach successfully outperforms a state-of-the-art confidence-based rule mining approach in mining Horn rules that involve compositional reasoning.

cs.CL↗

arXiv2015

A Deep Embedding Model for Co-occurrence Learning

Co-occurrence Data is a common and important information source in many areas, such as the word co-occurrence in the sentences, friends co-occurrence in social networks and products co-occurrence in commercial transaction data, etc, which contains rich correlation and clustering information about the items. In this paper, we study co-occurrence data using a general energy-based probabilistic model, and we analyze three different categories of energy-based model, namely, the $L_1$, $L_2$ and $L_k$ models, which are able to capture different levels of dependency in the co-occurrence data. We also discuss how several typical existing models are related to these three types of energy models, including the Fully Visible Boltzmann Machine (FVBM) ($L_2$), Matrix Factorization ($L_2$), Log-BiLinear (LBL) models ($L_2$), and the Restricted Boltzmann Machine (RBM) model ($L_k$). Then, we propose a Deep Embedding Model (DEM) (an $L_k$ model) from the energy model in a \emph{principled} manner. Furthermore, motivated by the observation that the partition function in the energy model is intractable and the fact that the major objective of modeling the co-occurrence data is to predict using the conditional probability, we apply the \emph{maximum pseudo-likelihood} method to learn DEM. In consequence, the developed model and its learning method naturally avoid the above difficulties and can be easily used to compute the conditional probability in prediction. Interestingly, our method is equivalent to learning a special structured deep neural network using back-propagation and a special sampling strategy, which makes it scalable on large-scale datasets. Finally, in the experiments, we show that the DEM can achieve comparable or better results than state-of-the-art methods on datasets across several application domains.

cs.LG↗

arXiv2015

Joint Learning of Distributed Representations for Images and Texts

This technical report provides extra details of the deep multimodal similarity model (DMSM) which was proposed in (Fang et al. 2015, arXiv:1411.4952). The model is trained via maximizing global semantic similarity between images and their captions in natural language using the public Microsoft COCO database, which consists of a large set of images and their corresponding captions. The learned representations attempt to capture the combination of various visual concepts and cues.

cs.CV↗

arXiv2015

From Captions to Visual Concepts and Back

This paper presents a novel approach for automatically generating image descriptions: visual detectors, language models, and multimodal similarity models learnt directly from a dataset of image captions. We use multiple instance learning to train visual detectors for words that commonly occur in captions, including many different parts of speech such as nouns, verbs, and adjectives. The word detector outputs serve as conditional inputs to a maximum-entropy language model. The language model learns from a set of over 400,000 image descriptions to capture the statistics of word usage. We capture global semantics by re-ranking caption candidates using sentence-level features and a deep multimodal similarity model. Our system is state-of-the-art on the official Microsoft COCO benchmark, producing a BLEU-4 score of 29.1%. When human judges compare the system captions to ones written by other people on our held-out test set, the system captions have equal or better quality 34% of the time.

cs.CV↗

arXiv2015

Intrinsic Defect Formation in Peptide Self-Assembly

In contrast to extensively studied defects in traditional materials, we report here for the first time a systematic investigation of the formation mechanism of intrinsic defects in self-assembled peptide nanostructures. The Monte Carlo simulations with our simplified dynamic hierarchical model revealed that the symmetry breaking of layer bending mode at the two ends during morphological transformation is responsible for intrinsic defect formation, whose microscopic origin is the mismatch between layer stacking along the side-chain direction and layer growth along the hydrogen bond direction. Moreover, defect formation does not affect the chirality of the self-assembled structure, which is determined by the initial steps of the peptide self-assembly process.

cond-mat.soft↗

arXiv2014

Learning Multi-Relational Semantics Using Neural-Embedding Models

In this paper we present a unified framework for modeling multi-relational representations, scoring, and learning, and conduct an empirical study of several recent multi-relational embedding models under the framework. We investigate the different choices of relation operators based on linear and bilinear transformations, and also the effects of entity representations by incorporating unsupervised vectors pre-trained on extra textual resources. Our results show several interesting findings, enabling the design of a simple embedding model that achieves the new state-of-the-art performance on a popular knowledge base completion task evaluated on Freebase.

cs.CL↗

arXiv2014

A Primal-Dual Method for Training Recurrent Neural Networks Constrained by the Echo-State Property

We present an architecture of a recurrent neural network (RNN) with a fully-connected deep neural network (DNN) as its feature extractor. The RNN is equipped with both causal temporal prediction and non-causal look-ahead, via auto-regression (AR) and moving-average (MA), respectively. The focus of this paper is a primal-dual training method that formulates the learning of the RNN as a formal optimization problem with an inequality constraint that provides a sufficient condition for the stability of the network dynamics. Experimental results demonstrate the effectiveness of this new method, which achieves 18.86% phone recognition error on the TIMIT benchmark for the core test set. The result approaches the best result of 17.7%, which was obtained by using RNN with long short-term memory (LSTM). The results also show that the proposed primal-dual training method produces lower recognition errors than the popular RNN methods developed earlier based on the carefully tuned threshold parameter that heuristically prevents the gradient from exploding.

cs.LG↗

arXiv2013

Learning Semantic Representations for the Phrase Translation Model

This paper presents a novel semantic-based phrase translation model. A pair of source and target phrases are projected into continuous-valued vector representations in a low-dimensional latent semantic space, where their translation score is computed by the distance between the pair in this new space. The projection is performed by a multi-layer neural network whose weights are learned on parallel training data. The learning is aimed to directly optimize the quality of end-to-end machine translation results. Experimental evaluation has been performed on two Europarl translation tasks, English-French and German-English. The results show that the new semantic-based phrase translation model significantly improves the performance of a state-of-the-art phrase-based statistical machine translation sys-tem, leading to a gain of 0.7-1.0 BLEU points.

cs.CL↗

arXiv2013

Learning Input and Recurrent Weight Matrices in Echo State Networks

Echo State Networks (ESNs) are a special type of the temporally deep network model, the Recurrent Neural Network (RNN), where the recurrent matrix is carefully designed and both the recurrent and input matrices are fixed. An ESN uses the linearity of the activation function of the output units to simplify the learning of the output matrix. In this paper, we devise a special technique that take advantage of this linearity in the output units of an ESN, to learn the input and recurrent matrices. This has not been done in earlier ESNs due to their well known difficulty in learning those matrices. Compared to the technique of BackPropagation Through Time (BPTT) in learning general RNNs, our proposed method exploits linearity of activation function in the output units to formulate the relationships amongst the various matrices in an RNN. These relationships results in the gradient of the cost function having an analytical form and being more accurate. This would enable us to compute the gradients instead of obtaining them by recursion as in BPTT. Experimental results on phone state classification show that learning one or both the input and recurrent matrices in an ESN yields superior results compared to traditional ESNs that do not learn these matrices, especially when longer time steps are used.

cs.LG↗

arXiv2012

Diffusion-limited aggregation with polygon particles

Diffusion-limited aggregation (DLA) assumes that particles perform pure random walk at a finite temperature and aggregate when they come close enough and stick together. Although it is well known that DLA in two dimensions results in a ramified fractal structure, how the particle shape influences the formed morphology is still unclear. In this work, we perform the off-lattice two-dimensional DLA simulations with different particle shapes of triangle, quadrangle, pentagon, hexagon, and octagon, respectively, and compared with the results for circular particles. Our results indicate that different particle shapes only change the local structure, but have no effects on the global structure of the formed fractal cluster. The local compactness decreases as the number of polygon edges increases.

cond-mat.stat-mech↗

arXiv2002

Time Domain Computation of a Nonlinear Nonlocal Cochlear Model with Applications to Multitone Interaction in Hearing

A nonlinear nonlocal cochlear model of the transmission line type is studied in order to capture the multitone interactions and resulting tonal suppression effects. The model can serve as a module for voice signal processing, it is a one dimensional (in space) damped dispersive nonlinear PDE based on mechanics and phenomenology of hearing. It describes the motion of basilar membrane (BM) in the cochlea driven by input pressure waves. Both elastic damping and selective longitudinal fluid damping are present. The former is nonlinear and nonlocal in BM displacement, and plays a key role in capturing tonal interactions. The latter is active only near the exit boundary (helicotrema), and is built in to damp out the remaining long waves. The initial boundary value problem is numerically solved with a semi-implicit second order finite difference method. Solutions reach a multi-frequency quasi-steady state. Numerical results are shown on two tone suppression from both high-frequency and low-frequency sides, consistent with known behavior of two tone suppression. Suppression effects among three tones are demonstrated by showing how the response magnitudes of the fixed two tones are reduced as we vary the third tone in frequency and amplitude. We observe qualitative agreement of our model solutions with existing cat auditory neural data. The model is thus simple and efficient as a processing tool for voice signals.

nlin.PS↗