arXiv subjects

Jing Pan

Publications and source records attributed to Jing Pan.

30 records · Page 2Linked to original sources

arXiv2021

Performance-Efficiency Trade-offs in Unsupervised Pre-training for Speech Recognition

This paper is a study of performance-efficiency trade-offs in pre-trained models for automatic speech recognition (ASR). We focus on wav2vec 2.0, and formalize several architecture designs that influence both the model performance and its efficiency. Putting together all our observations, we introduce SEW (Squeezed and Efficient Wav2vec), a pre-trained model architecture with significant improvements along both performance and efficiency dimensions across a variety of training setups. For example, under the 100h-960h semi-supervised setup on LibriSpeech, SEW achieves a 1.9x inference speedup compared to wav2vec 2.0, with a 13.5% relative reduction in word error rate. With a similar inference time, SEW reduces word error rate by 25-50% across different model sizes.

cs.CL↗

arXiv2021

A new ground-state band level energy formula for transitional nuclei

A new three parameter formula is proposed for ground-state bands in even-even soft rotors or called transitional nuclei. The new formula blends those of very soft nuclei and well deformed nuclei. Especially, it is found in fact that the formula proposed by P. von Brentano et al. can be derived from our formula if some approximations are adopted and it can be explained why in their work the energy-dependent term beta*E can be dropped. Compared to their formula it can be resulted in that the range of validity of our formula will be much broader than theirs.

nucl-th↗

arXiv2021

Deep-learning-based recognition of multi-singularity structured light

Structured light with customized complex topological pattern inspires diverse classical and quantum investigations underpinned by accurate detection techniques. However, the current detection schemes are limited to vortex beam with simple phase singularity. The precise recognition of general structured light with multiple singularities remains elusive. Here, we report a deep learning (DL) framework that can unveil multi-singularity phase structures in an end-to-end manner after feeding only two intensity patterns upon beam propagation captured via a camera, thus unleashing intuitive information of twisted photons. The DL toolbox can also acquire phases of Laguerre-Gaussian (LG) modes with single singularity and other general phase objects likewise. Leveraging this DL platform, a phase-based optical secret sharing (OSS) protocol is proposed, which is based on a more general class of multi-singularity modes than conventional LG beams. The OSS protocol features strong security, wealthy state space and convenient intensity-based measurements. This study opens new avenues for vortex beam communications, laser mode analysis, microscopy, Bose-Einstein condensates characterization, etc.

physics.optics↗

arXiv2021

Leveraging Pre-trained Language Model for Speech Sentiment Analysis

In this paper, we explore the use of pre-trained language models to learn sentiment information of written texts for speech sentiment analysis. First, we investigate how useful a pre-trained language model would be in a 2-step pipeline approach employing Automatic Speech Recognition (ASR) and transcripts-based sentiment analysis separately. Second, we propose a pseudo label-based semi-supervised training strategy using a language model on an end-to-end speech sentiment approach to take advantage of a large, but unlabeled speech dataset for training. Although spoken and written texts have different linguistic characteristics, they can complement each other in understanding sentiment. Therefore, the proposed system can not only model acoustic characteristics to bear sentiment-specific information in speech signals, but learn latent information to carry sentiments in the text representation. In these experiments, we demonstrate the proposed approaches improve F1 scores consistently compared to systems without a language model. Moreover, we also show that the proposed framework can reduce 65% of human supervision by leveraging a large amount of data without human sentiment annotation and boost performance in a low-resource condition where the human sentiment annotation is not available enough.

cs.CL↗

arXiv2021

Multistream CNN for Robust Acoustic Modeling

This paper proposes multistream CNN, a novel neural network architecture for robust acoustic modeling in speech recognition tasks. The proposed architecture processes input speech with diverse temporal resolutions by applying different dilation rates to convolutional neural networks across multiple streams to achieve the robustness. The dilation rates are selected from the multiples of a sub-sampling rate of 3 frames. Each stream stacks TDNN-F layers (a variant of 1D CNN), and output embedding vectors from the streams are concatenated then projected to the final layer. We validate the effectiveness of the proposed multistream CNN architecture by showing consistent improvements against Kaldi's best TDNN-F model across various data sets. Multistream CNN improves the WER of the test-other set in the LibriSpeech corpus by 12% (relative). On custom data from ASAPP's production ASR system for a contact center, it records a relative WER improvement of 11% for customer channel audio to prove its robustness to data in the wild. In terms of real-time factor, multistream CNN outperforms the baseline TDNN-F by 15%, which also suggests its practicality on production systems. When combined with self-attentive SRU LM rescoring, multistream CNN contributes for ASAPP to achieve the best WER of 1.75% on test-clean in LibriSpeech.

eess.AS↗

arXiv2020

Adversarial Validation Approach to Concept Drift Problem in User Targeting Automation Systems at Uber

In user targeting automation systems, concept drift in input data is one of the main challenges. It deteriorates model performance on new data over time. Previous research on concept drift mostly proposed model retraining after observing performance decreases. However, this approach is suboptimal because the system fixes the problem only after suffering from poor performance on new data. Here, we introduce an adversarial validation approach to concept drift problems in user targeting automation systems. With our approach, the system detects concept drift in new data before making inference, trains a model, and produces predictions adapted to the new data. We show that our approach addresses concept drift effectively with the AutoML3 Lifelong Machine Learning challenge data as well as in Uber's internal user targeting automation system, MaLTA.

cs.LG↗

arXiv2020

ASAPP-ASR: Multistream CNN and Self-Attentive SRU for SOTA Speech Recognition

In this paper we present state-of-the-art (SOTA) performance on the LibriSpeech corpus with two novel neural network architectures, a multistream CNN for acoustic modeling and a self-attentive simple recurrent unit (SRU) for language modeling. In the hybrid ASR framework, the multistream CNN acoustic model processes an input of speech frames in multiple parallel pipelines where each stream has a unique dilation rate for diversity. Trained with the SpecAugment data augmentation method, it achieves relative word error rate (WER) improvements of 4% on test-clean and 14% on test-other. We further improve the performance via N-best rescoring using a 24-layer self-attentive SRU language model, achieving WERs of 1.75% on test-clean and 4.46% on test-other.

eess.AS↗

arXiv2020

Benchmark Tests of Convolutional Neural Network and Graph Convolutional Network on HorovodRunner Enabled Spark Clusters

The freedom of fast iterations of distributed deep learning tasks is crucial for smaller companies to gain competitive advantages and market shares from big tech giants. HorovodRunner brings this process to relatively accessible spark clusters. There have been, however, no benchmark tests on HorovodRunner per se, nor specifically graph convolutional network (GCN, hereafter), and very limited scalability benchmark tests on Horovod, the predecessor requiring custom built GPU clusters. For the first time, we show that Databricks' HorovodRunner achieves significant lift in scaling efficiency for the convolutional neural network (CNN, hereafter) based tasks on both GPU and CPU clusters, but not the original GCN task. We also implemented the Rectified Adam optimizer for the first time in HorovodRunner.

cs.DC↗

arXiv2020

Two-dimensional-controlled high-order modes and vortex beams from an intracavity mode converter laser

We present a novel scheme of structured light laser with an astigmatic mode converter (AMC) as intracavity element, first enabling the generation of Hermite-Gaussian (HG) modes with fully controlled two-dimensional (2D) indices (m,n) and vortex beams carrying orbital angular momentum (OAM) directly from cavity. The 2D tunability was realized by controlling the off-axis displacements of both pump and intracavity AMC. The output HGm,n beam could be externally converted into OAM beam with 2D tunable radial and azimuthal indices (p,l). With the certain parameter control, vortex beam carrying OAM also could be directly generated from the cavity. Our setup provides a compact and concise structured light source. It has great potential in extending various applications of optical tweezers, communications, and nonlinearity.

physics.optics↗

arXiv2019

Order Matters at Fanatics Recommending Sequentially Ordered Products by LSTM Embedded with Word2Vec

A unique challenge for e-commerce recommendation is that customers are often interested in products that are more advanced than their already purchased products, but not reversed. The few existing recommender systems modeling unidirectional sequence output a limited number of categories or continuous variables. To model the ordered sequence, we design the first recommendation system that both embed purchased items with Word2Vec, and model the sequence with stateless LSTM RNN. The click-through rate of this recommender system in production outperforms its solely Word2Vec based predecessor. Developed in 2017, it was perhaps the first published real-world application that makes distributed predictions of a single machine trained Keras model on Spark slave nodes at a scale of more than 0.4 million columns per row.

cs.LG↗

arXiv2018

Ultrasensitive hybrid optical skin

Electronic skin, a class of wearable electronic sensors that mimic the functionalities of human skin, has made remarkable success in applications including health monitoring, human-machine interaction and electronic-biological interfaces. While electronic skin continues to achieve higher sensitivity and faster response, its ultimate performance is fundamentally limited by the nature of low-frequency AC currents in electronic circuitries. Here we demonstrate highly sensitive optical skin (O-skin) in which the primary sensory elements are optically driven. The simple construction of the sensors is achieved by embedding glass micro/nanofibers (MNFs) in thin layers of polydimethylsiloxane (PDMS). Enabled by the highly sensitive power-leakage response of the guided modes from the MNF upon external stimuli, our optical sensors show ultrahigh sensitivity (1870/kPa), low detection limit (7 mPa) and fast response (10 microseconds) for pressure sensing, significantly exceeding the performance metrics of state-of-the-art electronic skins. Electromagnetic interference (EMI)-free detection of high-frequency vibrations, wrist pulse and human voice are realized. Moreover, a five-sensor optical data glove and a 2x2-MNF tactile sensor are demonstrated. Our results pave the way toward wearable optical devices ranging from ultrasensitive flexible sensors to optical skins.

physics.app-ph↗

arXiv2015

Moving Object Detection in Video Using Saliency Map and Subspace Learning

Moving object detection is a key to intelligent video analysis. On the one hand, what moves is not only interesting objects but also noise and cluttered background. On the other hand, moving objects without rich texture are prone not to be detected. So there are undesirable false alarms and missed alarms in many algorithms of moving object detection. To reduce the false alarms and missed alarms, in this paper, we propose to incorporate a saliency map into an incremental subspace analysis framework where the saliency map makes estimated background has less chance than foreground (i.e., moving objects) to contain salient objects. The proposed objective function systematically takes account into the properties of sparsity, low-rank, connectivity, and saliency. An alternative minimization algorithm is proposed to seek the optimal solutions. Experimental results on the Perception Test Images Sequences demonstrate that the proposed method is effective in reducing false alarms and missed alarms.

cs.CV↗