arXiv subjects

Bo Xu

Publications and source records attributed to Bo Xu.

At least 271 records · Page 15Linked to original sources

arXiv2019

Extending Recurrent Neural Aligner for Streaming End-to-End Speech Recognition in Mandarin

End-to-end models have been showing superiority in Automatic Speech Recognition (ASR). At the same time, the capacity of streaming recognition has become a growing requirement for end-to-end models. Following these trends, an encoder-decoder recurrent neural network called Recurrent Neural Aligner (RNA) has been freshly proposed and shown its competitiveness on two English ASR tasks. However, it is not clear if RNA can be further improved and applied to other spoken language. In this work, we explore the applicability of RNA in Mandarin Chinese and present four effective extensions: In the encoder, we redesign the temporal down-sampling and introduce a powerful convolutional structure. In the decoder, we utilize a regularizer to smooth the output distribution and conduct joint training with a language model. On two Mandarin Chinese conversational telephone speech recognition (MTS) datasets, our Extended-RNA obtains promising performance. Particularly, it achieves 27.7% character error rate (CER), which is superior to current state-of-the-art result on the popular HKUST task.

cs.SD↗

arXiv2019

Self-Attention Aligner: A Latency-Control End-to-End Model for ASR Using Self-Attention Network and Chunk-Hopping

Self-attention network, an attention-based feedforward neural network, has recently shown the potential to replace recurrent neural networks (RNNs) in a variety of NLP tasks. However, it is not clear if the self-attention network could be a good alternative of RNNs in automatic speech recognition (ASR), which processes the longer speech sequences and may have online recognition requirements. In this paper, we present a RNN-free end-to-end model: self-attention aligner (SAA), which applies the self-attention networks to a simplified recurrent neural aligner (RNA) framework. We also propose a chunk-hopping mechanism, which enables the SAA model to encode on segmented frame chunks one after another to support online recognition. Experiments on two Mandarin ASR datasets show the replacement of RNNs by the self-attention networks yields a 8.4%-10.2% relative character error rate (CER) reduction. In addition, the chunk-hopping mechanism allows the SAA to have only a 2.5% relative CER degradation with a 320ms latency. After jointly training with a self-attention network language model, our SAA model obtains further error rate reduction on multiple datasets. Especially, it achieves 24.12% CER on the Mandarin ASR benchmark (HKUST), exceeding the best end-to-end model by over 2% absolute CER.

cs.CL↗

arXiv2018

Concept Learning through Deep Reinforcement Learning with Memory-Augmented Neural Networks

Deep neural networks have shown superior performance in many regimes to remember familiar patterns with large amounts of data. However, the standard supervised deep learning paradigm is still limited when facing the need to learn new concepts efficiently from scarce data. In this paper, we present a memory-augmented neural network which is motivated by the process of human concept learning. The training procedure, imitating the concept formation course of human, learns how to distinguish samples from different classes and aggregate samples of the same kind. In order to better utilize the advantages originated from the human behavior, we propose a sequential process, during which the network should decide how to remember each sample at every step. In this sequential process, a stable and interactive memory serves as an important module. We validate our model in some typical one-shot learning tasks and also an exploratory outlier detection problem. In all the experiments, our model gets highly competitive to reach or outperform those strong baselines.

cs.LG↗

arXiv2018

Block Chain based Intelligent Industrial Network (DSDIN)

The manufacturing industry featured centralization in the past due to technical limitations, and factories (especially large manufacturers) gathered almost all of the resources for manufacturing, including: technologies, raw materials, equipment, workers, market information, etc. However, such centralized production is costly, inefficient and inflexible, and difficult to respond to rapidly changing, diverse and personalized user needs. This paper introduces an Intelligent Industrial Network (DSDIN), which provides a fully distributed manufacturing network where everyone can participate in manufacturing due to decentralization and no intermediate links, allowing them to quickly get the products or services they want and also to be authorized, recognized and get returns in a low-cost way due to their efforts (such as providing creative ideas, designs or equipment, raw materials or physical strength). DSDIN is a blockchain based IoT and AI technology platform, and also an IoT based intelligent service standard. Due to the intelligent network formed by DSDIN, the manufacturing center is no longer a factory, and actually there are no manufacturing centers. DSDIN provides a multi-participation peer-to-peer network for people and things (including raw materials, equipment, finished / semi-finished products, etc.). The information transmitted through the network is called Intelligent Service Algorithm (ISA). The user can send a process model, formula or control parameter to a device via an ISA, and every transaction in DSDIN is an intelligent service defined by ISA.

cs.CY↗

arXiv2018

Cascaded Mutual Modulation for Visual Reasoning

Visual reasoning is a special visual question answering problem that is multi-step and compositional by nature, and also requires intensive text-vision interactions. We propose CMM: Cascaded Mutual Modulation as a novel end-to-end visual reasoning model. CMM includes a multi-step comprehension process for both question and image. In each step, we use a Feature-wise Linear Modulation (FiLM) technique to enable textual/visual pipeline to mutually control each other. Experiments show that CMM significantly outperforms most related models, and reach state-of-the-arts on two visual reasoning benchmarks: CLEVR and NLVR, collected from both synthetic and natural languages. Ablation studies confirm that both our multistep framework and our visual-guided language modulation are critical to the task. Our code is available at https://github.com/FlamingHorizon/CMM-VR.

cs.IR↗

arXiv2018

Electronic quantum coherence induced by strong field molecular ionization

The existence of electronic coherence can fundamentally change the scenario of nonlinear interaction of light with quantum systems such as atoms and molecules, which, however, has escaped from observation in the investigations of strong field nonlinear optics in the past several decades. Here, we report on the generation of electronic quantum coherence by strong field ionization of nitrogen molecules in an intense 800 nm laser field. The coherence is experimentally revealed by observing a resonant four-wave mixing process in which the two pump pulses centered at 800 nm and 1580 nm wavelengths are temporally separated from each other. The experimental observation is further reproduced by calculating the nonlinear polarization response of N_2^+ ions using a three-level quantum model. Our result suggests that strong field ionization provides a unique approach to generating a fully coherent molecular wavepacket encapsulating the rotational, vibrational, and electronic states.

physics.optics↗

arXiv2018

Single-channel Speech Dereverberation via Generative Adversarial Training

In this paper, we propose a single-channel speech dereverberation system (DeReGAT) based on convolutional, bidirectional long short-term memory and deep feed-forward neural network (CBLDNN) with generative adversarial training (GAT). In order to obtain better speech quality instead of only minimizing a mean square error (MSE), GAT is employed to make the dereverberated speech indistinguishable form the clean samples. Besides, our system can deal with wide range reverberation and be well adapted to variant environments. The experimental results show that the proposed model outperforms weighted prediction error (WPE) and deep neural network-based systems. In addition, DeReGAT is extended to an online speech dereverberation scenario, which reports comparable performance with the offline case.

cs.SD↗

arXiv2018

Multilingual End-to-End Speech Recognition with A Single Transformer on Low-Resource Languages

Sequence-to-sequence attention-based models integrate an acoustic, pronunciation and language model into a single neural network, which make them very suitable for multilingual automatic speech recognition (ASR). In this paper, we are concerned with multilingual speech recognition on low-resource languages by a single Transformer, one of sequence-to-sequence attention-based models. Sub-words are employed as the multilingual modeling unit without using any pronunciation lexicon. First, we show that a single multilingual ASR Transformer performs well on low-resource languages despite of some language confusion. We then look at incorporating language information into the model by inserting the language symbol at the beginning or at the end of the original sub-words sequence under the condition of language information being known during training. Experiments on CALLHOME datasets demonstrate that the multilingual ASR Transformer with the language symbol at the end performs better and can obtain relatively 10.5\% average word error rate (WER) reduction compared to SHL-MLSTM with residual learning. We go on to show that, assuming the language information being known during training and testing, about relatively 12.4\% average WER reduction can be observed compared to SHL-MLSTM with residual learning through giving the language symbol as the sentence start token.

eess.AS↗

arXiv2018

Syllable-Based Sequence-to-Sequence Speech Recognition with the Transformer in Mandarin Chinese

Sequence-to-sequence attention-based models have recently shown very promising results on automatic speech recognition (ASR) tasks, which integrate an acoustic, pronunciation and language model into a single neural network. In these models, the Transformer, a new sequence-to-sequence attention-based model relying entirely on self-attention without using RNNs or convolutions, achieves a new single-model state-of-the-art BLEU on neural machine translation (NMT) tasks. Since the outstanding performance of the Transformer, we extend it to speech and concentrate on it as the basic architecture of sequence-to-sequence attention-based model on Mandarin Chinese ASR tasks. Furthermore, we investigate a comparison between syllable based model and context-independent phoneme (CI-phoneme) based model with the Transformer in Mandarin Chinese. Additionally, a greedy cascading decoder with the Transformer is proposed for mapping CI-phoneme sequences and syllable sequences into word sequences. Experiments on HKUST datasets demonstrate that syllable based model with the Transformer performs better than CI-phoneme based counterpart, and achieves a character error rate (CER) of \emph{$28.77\%$}, which is competitive to the state-of-the-art CER of $28.0\%$ by the joint CTC-attention based encoder-decoder network.

eess.AS↗

arXiv2018

Low-noise 750-MHz spaced Yb:fiber frequency combs

We demonstrate two fully and tightly phase locked 750 MHz ytterbium (Yb) fiber frequency combs that are independently stabilized to a continuous wave (CW) laser with <1 rad RMS phase error. A bulk EOM and a single stack PZT are separately utilized as the fast actuators for cavity length stabilization. The carrier envelop frequencies are phase locked by single loop feedback to laser diode current, showing 1.6 MHz servo bumps. The in-loop fractional frequency instabilities are ~1.5e-18 at 1s for both combs. To the best of our knowledge, this is the highest repetition rate in fiber based low phase noise combs tightly locked to optical frequency reference.

physics.ins-det↗

arXiv2018

A Comparison of Modeling Units in Sequence-to-Sequence Speech Recognition with the Transformer on Mandarin Chinese

The choice of modeling units is critical to automatic speech recognition (ASR) tasks. Conventional ASR systems typically choose context-dependent states (CD-states) or context-dependent phonemes (CD-phonemes) as their modeling units. However, it has been challenged by sequence-to-sequence attention-based models, which integrate an acoustic, pronunciation and language model into a single neural network. On English ASR tasks, previous attempts have already shown that the modeling unit of graphemes can outperform that of phonemes by sequence-to-sequence attention-based model. In this paper, we are concerned with modeling units on Mandarin Chinese ASR tasks using sequence-to-sequence attention-based models with the Transformer. Five modeling units are explored including context-independent phonemes (CI-phonemes), syllables, words, sub-words and characters. Experiments on HKUST datasets demonstrate that the lexicon free modeling units can outperform lexicon related modeling units in terms of character error rate (CER). Among five modeling units, character based model performs best and establishes a new state-of-the-art CER of $26.64\%$ on HKUST datasets without a hand-designed lexicon and an extra language model integration, which corresponds to a $4.8\%$ relative improvement over the existing best CER of $28.0\%$ by the joint CTC-attention based encoder-decoder network.

eess.AS↗

arXiv2018

Terabit optical OFDM superchannel transmission via coherent carriers of a hybrid chip-scale soliton frequency comb

We demonstrate seamless channel multiplexing and high bitrate superchannel transmission of coherent optical orthogonal-frequency-division-multiplexing (CO-OFDM) data signals utilizing a dissipative Kerr soliton (DKS) frequency comb generated in an on-chip microcavity. Aided by comb line multiplication through Nyquist pulse modulation, the high stability and mutual coherence among mode-locked Kerr comb lines are exploited for the first time to eliminate the guard intervals between communication channels and achieve full spectral density bandwidth utilization. Spectral efficiency as high as 2.625 bit/Hz/s is obtained for 180 CO-OFDM bands encoded with 12.75 Gbaud 8-QAM data, adding up to total bitrate of 6.885 Tb/s within 2.295 THz frequency comb bandwidth. Our study confirms that high coherence is the key superiority of Kerr soliton frequency combs over independent laser diodes, as a multi-spectral coherent laser source for high-bandwidth high-spectral-density transmission networks.

physics.app-ph↗

arXiv2018

Unsupervised Neural Machine Translation with Weight Sharing

Unsupervised neural machine translation (NMT) is a recently proposed approach for machine translation which aims to train the model without using any labeled data. The models proposed for unsupervised NMT often use only one shared encoder to map the pairs of sentences from different languages to a shared-latent space, which is weak in keeping the unique and internal characteristics of each language, such as the style, terminology, and sentence structure. To address this issue, we introduce an extension by utilizing two independent encoders but sharing some partial weights which are responsible for extracting high-level representations of the input sentences. Besides, two different generative adversarial networks (GANs), namely the local GAN and global GAN, are proposed to enhance the cross-language translation. With this new approach, we achieve significant improvements on English-German, English-French and Chinese-to-English translation tasks.

cs.CL↗

arXiv2018

Improving Neural Machine Translation with Conditional Sequence Generative Adversarial Nets

This paper proposes an approach for applying GANs to NMT. We build a conditional sequence generative adversarial net which comprises of two adversarial sub models, a generator and a discriminator. The generator aims to generate sentences which are hard to be discriminated from human-translated sentences (i.e., the golden target sentences), And the discriminator makes efforts to discriminate the machine-generated sentences from human-translated ones. The two sub models play a mini-max game and achieve the win-win situation when they reach a Nash Equilibrium. Additionally, the static sentence-level BLEU is utilized as the reinforced objective for the generator, which biases the generation towards high BLEU points. During training, both the dynamic discriminator and the static BLEU objective are employed to evaluate the generated sentences and feedback the evaluations to guide the learning of the generator. Experimental results show that the proposed model consistently outperforms the traditional RNNSearch and the newly emerged state-of-the-art Transformer on English-German and Chinese-English translation tasks.

cs.CL↗

arXiv2018

On existence of out-of-plane equilibrium points in restricted three-body problem with oblateness

We analyze in this paper the existence of the "out-of-plane" equilibrium points in the restricted three-body problem with oblateness. From the series expansion of the potential function of an oblate asteroid, we show analytically all equilibrium points locate on the orbital plane of primaries and how artificial equilibrium points may arise due to an inappropriate application of the potential function. Using the closed form of the potential of a triaxial ellipsoid, we analytically demonstrate that the gravitational acceleration in $z$-direction is always pointing toward the equatorial plane, thus it could not be balanced out at any value of $z\neq 0$ and the out-of-plane equilibrium points cannot exist. The out-of-plane equilibrium points appear only when additional acceleration other than the gravitation from primaries is taken into account. We suggest that special attention must be paid to the application of the spherical harmonics expansion of potential to find the equilibrium points, especially when these points may be very close to the celestial body.

astro-ph.EP↗

arXiv2017

Convolutional Neural Network with Word Embeddings for Chinese Word Segmentation

Character-based sequence labeling framework is flexible and efficient for Chinese word segmentation (CWS). Recently, many character-based neural models have been applied to CWS. While they obtain good performance, they have two obvious weaknesses. The first is that they heavily rely on manually designed bigram feature, i.e. they are not good at capturing n-gram features automatically. The second is that they make no use of full word information. For the first weakness, we propose a convolutional neural model, which is able to capture rich n-gram features without any feature engineering. For the second one, we propose an effective approach to integrate the proposed model with word embeddings. We evaluate the model on two benchmark datasets: PKU and MSR. Without any feature engineering, the model obtains competitive performance -- 95.7% on PKU and 97.3% on MSR. Armed with word embeddings, the model achieves state-of-the-art performance on both datasets -- 96.5% on PKU and 98.0% on MSR, without using any external labeled resource.

cs.CL↗

arXiv2017

Near-resonant Raman amplification in the rotational quantum wavepackets of nitrogen molecular ions generated by strong field ionization

Generation of laser-like narrow bandwidth emissions from nitrogen molecular ions generated in intense near- and mid-infrared femtosecond laser fields has aroused much interest because of the mysterious physics underlying such a phenomenon as well as the potential application of such an effect in atmospheric spectroscopic sensing. Here, we perform a pump-probe measurement on the nonlinear interaction of rotational quantum wavepackets of nitrogen molecular ions generated in mid-infrared (e.g., at a wavelength centered at 1580 nm) femtosecond laser fields with an ultrashort probe pulse whose broad spectrum overlaps both P- and R-branch rotational transition lines between the upper and lower electronic states. The results show that in the near-resonant conditions, stimulated Raman amplification can efficiently occur which converts the broad bandwidth ultrashort probe pulse into the narrow bandwidth laser-like beam. Our finding provides an insight into the physical mechanism of strong field induced lasing actions in atmospheric environment.

physics.optics↗

arXiv2017

Generation of bright laser-like coherent emissions from nitrogen molecular ions in intense infrared laser fields: A unique quantum system for extreme nonlinear optics

We report on an investigation of simultaneous generation of several narrow-bandwidth laser-like coherent emissions from nitrogen molecular ions ( ) produced in intense mid-infrared laser fields. With systematic examinations on the dependences of coherent emissions on gas pressure as well as several laser parameters including laser intensity, polarization and wavelength of the pump laser pulses, we reveal that the multiple coherent emission lines generated in originate from a series of nonlinear processes beginning with four-wave mixing, followed with stimulated Raman scattering. Our analyses further show that the observed nonlinear processes are greatly enhanced at the resonant wavelengths, giving rise to high conversion efficiencies from the infrared pump laser pulses to the coherent emission lines near the transition wavelengths between the different vibrational energy levels of ground X and that of the excited B states.

physics.optics↗