arXiv subjects

Ke Hu

Publications and source records attributed to Ke Hu.

At least 55 records · Page 3Linked to original sources

arXiv2023

Mixture-of-Expert Conformer for Streaming Multilingual ASR

End-to-end models with large capacity have significantly improved multilingual automatic speech recognition, but their computation cost poses challenges for on-device applications. We propose a streaming truly multilingual Conformer incorporating mixture-of-expert (MoE) layers that learn to only activate a subset of parameters in training and inference. The MoE layer consists of a softmax gate which chooses the best two experts among many in forward propagation. The proposed MoE layer offers efficient inference by activating a fixed number of parameters as the number of experts increases. We evaluate the proposed model on a set of 12 languages, and achieve an average 11.9% relative improvement in WER over the baseline. Compared to an adapter model using ground truth information, our MoE model achieves similar WER and activates similar number of parameters but without any language information. We further show around 3% relative WER improvement by multilingual shallow fusion.

cs.CL↗

arXiv2023

A Deliberation-based Joint Acoustic and Text Decoder

We propose a new two-pass E2E speech recognition model that improves ASR performance by training on a combination of paired data and unpaired text data. Previously, the joint acoustic and text decoder (JATD) has shown promising results through the use of text data during model training and the recently introduced deliberation architecture has reduced recognition errors by leveraging first-pass decoding results. Our method, dubbed Deliberation-JATD, combines the spelling correcting abilities of deliberation with JATD's use of unpaired text data to further improve performance. The proposed model produces substantial gains across multiple test sets, especially those focused on rare words, where it reduces word error rate (WER) by between 12% and 22.5% relative. This is done without increasing model size or requiring multi-stage training, making Deliberation-JATD an efficient candidate for on-device applications.

eess.AS↗

arXiv2023

Massively Multilingual Shallow Fusion with Large Language Models

While large language models (LLM) have made impressive progress in natural language processing, it remains unclear how to utilize them in improving automatic speech recognition (ASR). In this work, we propose to train a single multilingual language model (LM) for shallow fusion in multiple languages. We push the limits of the multilingual LM to cover up to 84 languages by scaling up using a mixture-of-experts LLM, i.e., generalist language model (GLaM). When the number of experts increases, GLaM dynamically selects only two at each decoding step to keep the inference computation roughly constant. We then apply GLaM to a multilingual shallow fusion task based on a state-of-the-art end-to-end model. Compared to a dense LM of similar computation during inference, GLaM reduces the WER of an English long-tail test set by 4.4% relative. In a multilingual shallow fusion task, GLaM improves 41 out of 50 languages with an average relative WER reduction of 3.85%, and a maximum reduction of 10%. Compared to the baseline model, GLaM achieves an average WER reduction of 5.53% over 43 languages.

cs.CL↗

arXiv2023

High-efficient harmonic vortex generation from a laser irradiated hollow-cone target

It has been recently reported that ultraviolet harmonic vortices can be produced when a high-power circular-polarized laser pulse travels through a micro-scale waveguide. However, the harmonic generation quenches typically after a few tens of microns of propagation, due to the build-up of electrostatic potential that suppresses the amplitude of the surface wave. Here we propose to utilize a hollow-cone channel to overcome this obstacle. When traveling in a cone target, the laser intensity gradually increases, allowing the surface wave to maintain a high amplitude for a much longer distance. The harmonic vortices can be produced with very high efficiency. According to three-dimensional particle-in-cell simulations, the overall efficiency are boosted by almost one order of magnitude, reaching $>20\%$. It is also found that using the cone targets can mitigate efficiency decline due to pre-heating of plasma by laser prepulse. The proposed scheme paves the way to the development of powerful optical vortices sources in the extreme ultraviolet regime - an area of significant fundamental and applied physics potential.

physics.plasm-ph↗

arXiv2022

Hot subdwarf stars identified in LAMOST DR8 with single-lined and composite spectra

222 hot subdwarf stars were identified with LAMOST DR8 spectra, among which 131 stars show composite spectra and have been decomposed, while 91 stars present single-lined spectra. Atmospheric parameters of all sample stars were obtained by fitting Hydrogen (H) and Helium (He) line profiles with synthetic spectra. Two long-period composite sdB binaries were newly discovered by combining our sample with the non-single star data from Gaia DR3. One of the new systems presents the highest eccentricity (i.e., 0.5 +/- 0.09) among known wide sdB binaries, which is beyond model predictions. 15 composite sdB stars fall in the high probability binary region of RUWE-AEN plane, and deserve priority follow-up observations to further study their binary nature. A distinct gap is clearly presented among temperatures of cool companions for our composite-spectra sample. But we could not come to a conclusion whether this feature is connected to the formation history of hot subdwarf stars before their binary natures are confirmed.

astro-ph.SR↗

arXiv2022

Scaling Up Deliberation for Multilingual ASR

Multilingual end-to-end automatic speech recognition models are attractive due to its simplicity in training and deployment. Recent work on large-scale training of such models has shown promising results compared to monolingual models. However, the work often focuses on multilingual models themselves in a single-pass setup. In this work, we investigate second-pass deliberation for multilingual speech recognition. Our proposed deliberation is multilingual, i.e., the text encoder encodes hypothesis text from multiple languages, and the decoder attends to multilingual text and audio. We investigate scaling the deliberation text encoder and decoder, and compare scaling the deliberation decoder and the first-pass cascaded encoder. We show that deliberation improves the average WER on 9 languages by 4% relative compared to the single-pass model. By increasing the size of the deliberation up to 1B parameters, the average WER improvement increases to 9%, with up to 14% for certain languages. Our deliberation rescorer is based on transformer layers and can be parallelized during rescoring.

cs.CL↗

arXiv2022

Improving Deliberation by Text-Only and Semi-Supervised Training

Text-only and semi-supervised training based on audio-only data has gained popularity recently due to the wide availability of unlabeled text and speech data. In this work, we propose incorporating text-only and semi-supervised training into an attention-based deliberation model. By incorporating text-only data in training a bidirectional encoder representation from transformer (BERT) for the deliberation text encoder, and large-scale text-to-speech and audio-only utterances using joint acoustic and text decoder (JATD) and semi-supervised training, we achieved 4%-12% WER reduction for various tasks compared to the baseline deliberation. Compared to a state-of-the-art language model (LM) rescoring method, the deliberation model reduces the Google Voice Search WER by 11% relative. We show that the deliberation model also achieves a positive human side-by-side evaluation compared to the state-of-the-art LM rescorer with reasonable endpointer latencies.

cs.CL↗

arXiv2022

Hybrid CNN Based Attention with Category Prior for User Image Behavior Modeling

User historical behaviors are proved useful for Click Through Rate (CTR) prediction in online advertising system. In Meituan, one of the largest e-commerce platform in China, an item is typically displayed with its image and whether a user clicks the item or not is usually influenced by its image, which implies that user's image behaviors are helpful for understanding user's visual preference and improving the accuracy of CTR prediction. Existing user image behavior models typically use a two-stage architecture, which extracts visual embeddings of images through off-the-shelf Convolutional Neural Networks (CNNs) in the first stage, and then jointly trains a CTR model with those visual embeddings and non-visual features. We find that the two-stage architecture is sub-optimal for CTR prediction. Meanwhile, precisely labeled categories in online ad systems contain abundant visual prior information, which can enhance the modeling of user image behaviors. However, off-the-shelf CNNs without category prior may extract category unrelated features, limiting CNN's expression ability. To address the two issues, we propose a hybrid CNN based attention module, unifying user's image behaviors and category prior, for CTR prediction. Our approach achieves significant improvements in both online and offline experiments on a billion scale real serving dataset.

cs.CV↗

arXiv2022

Streaming Align-Refine for Non-autoregressive Deliberation

We propose a streaming non-autoregressive (non-AR) decoding algorithm to deliberate the hypothesis alignment of a streaming RNN-T model. Our algorithm facilitates a simple greedy decoding procedure, and at the same time is capable of producing the decoding result at each frame with limited right context, thus enjoying both high efficiency and low latency. These advantages are achieved by converting the offline Align-Refine algorithm to be streaming-compatible, with a novel transformer decoder architecture that performs local self-attentions for both text and audio, and a time-aligned cross-attention at each layer. Furthermore, we perform discriminative training of our model with the minimum word error rate (MWER) criterion, which has not been done in the non-AR decoding literature. Experiments on voice search datasets and Librispeech show that with reasonable right context, our streaming model performs as well as the offline counterpart, and discriminative training leads to further WER gain when the first-pass model has small capacity.

cs.CL↗

arXiv2022

Intense high-harmonic optical vortices generated from a micro-plasma-waveguide irradiated by a circularly polarized laser pulse

A scheme for generating intense high-harmonic optical vortices is proposed. It relies on spin-orbit interaction of light when a relativistically-strong circularly polarized laser pulse irradiates a micro-plasma-waveguide. The intense laser field drives a strong surface wave at the inner boundary of the waveguide, which leads to high-order harmonic generation as the laser traveling inside. For a circularly polarized drive laser, the optical chirality is imprinted to the surface wave, which facilitates conversion of spin angular momentum of the fundamental light into orbital angular momenta of the harmonics. A "shaken waveguide" model is developed showing that the aforementioned phenomena arises due to nonlinear plasma response that modifies electromagnetic mode at high intensities. We further show the phase velocities of all the harmonic beams are automatically matched to the driving laser, so that the harmonic intensities increase with propagation distance. The efficiency of harmonic production are related to the surface wave breaking effect, which can be significantly enhanced using a tightly focused laser. Our simulation suggests an overall conversion efficiency $\sim5\%$ can be achieved.

physics.plasm-ph↗

arXiv2022

Continual Learning for CTR Prediction: A Hybrid Approach

Click-through rate(CTR) prediction is a core task in cost-per-click(CPC) advertising systems and has been studied extensively by machine learning practitioners. While many existing methods have been successfully deployed in practice, most of them are built upon i.i.d.(independent and identically distributed) assumption, ignoring that the click data used for training and inference is collected through time and is intrinsically non-stationary and drifting. This mismatch will inevitably lead to sub-optimal performance. To address this problem, we formulate CTR prediction as a continual learning task and propose COLF, a hybrid COntinual Learning Framework for CTR prediction, which has a memory-based modular architecture that is designed to adapt, learn and give predictions continuously when faced with non-stationary drifting click data streams. Married with a memory population method that explicitly controls the discrepancy between memory and target data, COLF is able to gain positive knowledge from its historical experience and makes improved CTR predictions. Empirical evaluations on click log collected from a major shopping app in China demonstrate our method's superiority over existing methods. Additionally, we have deployed our method online and observed significant CTR and revenue improvement, which further demonstrates our method's efficacy.

cs.IR↗

arXiv2021

Deliberation of Streaming RNN-Transducer by Non-autoregressive Decoding

We propose to deliberate the hypothesis alignment of a streaming RNN-T model with the previously proposed Align-Refine non-autoregressive decoding method and its improved versions. The method performs a few refinement steps, where each step shares a transformer decoder that attends to both text features (extracted from alignments) and audio features, and outputs complete updated alignments. The transformer decoder is trained with the CTC loss which facilitates parallel greedy decoding, and performs full-context attention to capture label dependencies. We improve Align-Refine by introducing cascaded encoder that captures more audio context before refinement, and alignment augmentation which enforces learning label dependency. We show that, conditioned on hypothesis alignments of a streaming RNN-T model, our method obtains significantly more accurate recognition results than the first-pass RNN-T, with only small amount of model parameters.

cs.CL↗

arXiv2021

AutoHEnsGNN: Winning Solution to AutoGraph Challenge for KDD Cup 2020

Graph Neural Networks (GNNs) have become increasingly popular and achieved impressive results in many graph-based applications. However, extensive manual work and domain knowledge are required to design effective architectures, and the results of GNN models have high variance with different training setups, which limits the application of existing GNN models. In this paper, we present AutoHEnsGNN, a framework to build effective and robust models for graph tasks without any human intervention. AutoHEnsGNN won first place in the AutoGraph Challenge for KDD Cup 2020, and achieved the best rank score of five real-life datasets in the final phase. Given a task, AutoHEnsGNN first applies a fast proxy evaluation to automatically select a pool of promising GNN models. Then it builds a hierarchical ensemble framework: 1) We propose graph self-ensemble (GSE), which can reduce the variance of weight initialization and efficiently exploit the information of local and global neighborhoods; 2) Based on GSE, a weighted ensemble of different types of GNN models is used to effectively learn more discriminative node representations. To efficiently search the architectures and ensemble weights, we propose AutoHEnsGNN$_{\text{Gradient}}$, which treats the architectures and ensemble weights as architecture parameters and uses gradient-based architecture search to obtain optimal configurations, and AutoHEnsGNN$_{\text{Adaptive}}$, which can adaptively adjust the ensemble weight based on the model accuracy. Extensive experiments on node classification, graph classification, edge prediction and KDD Cup challenge demonstrate the effectiveness and generality of AutoHEnsGNN

cs.LG↗

arXiv2021

Polarized skylight orientation determination artificial neural network

This paper proposes an artificial neural network to determine orientation using polarized skylight. This neural network has specific dilated convolution, which can extract light intensity information of different polarization directions. Then, the degree of polarization (DOP) and angle of polarization (AOP) are directly extracted in the network. In addition, the exponential function encoding of orientation is designed as the network output, which can better reflect the insect's encoding of polarization information, and improve the accuracy of orientation determination. Finally, training and testing were conducted on a public polarized skylight navigation dataset, and the experimental results proved the stability and effectiveness of the network.

cs.CV↗

arXiv2021

Deep Position-wise Interaction Network for CTR Prediction

Click-through rate (CTR) prediction plays an important role in online advertising and recommender systems. In practice, the training of CTR models depends on click data which is intrinsically biased towards higher positions since higher position has higher CTR by nature. Existing methods such as actual position training with fixed position inference and inverse propensity weighted training with no position inference alleviate the bias problem to some extend. However, the different treatment of position information between training and inference will inevitably lead to inconsistency and sub-optimal online performance. Meanwhile, the basic assumption of these methods, i.e., the click probability is the product of examination probability and relevance probability, is oversimplified and insufficient to model the rich interaction between position and other information. In this paper, we propose a Deep Position-wise Interaction Network (DPIN) to efficiently combine all candidate items and positions for estimating CTR at each position, achieving consistency between offline and online as well as modeling the deep non-linear interaction among position, user, context and item under the limit of serving performance. Following our new treatment to the position bias in CTR prediction, we propose a new evaluation metrics named PAUC (position-wise AUC) that is suitable for measuring the ranking quality at a given position. Through extensive experiments on a real world dataset, we show empirically that our method is both effective and efficient in solving position bias problem. We have also deployed our method in production and observed statistically significant improvement over a highly optimized baseline in a rigorous A/B test.

cs.IR↗

arXiv2021

Learning Word-Level Confidence For Subword End-to-End ASR

We study the problem of word-level confidence estimation in subword-based end-to-end (E2E) models for automatic speech recognition (ASR). Although prior works have proposed training auxiliary confidence models for ASR systems, they do not extend naturally to systems that operate on word-pieces (WP) as their vocabulary. In particular, ground truth WP correctness labels are needed for training confidence models, but the non-unique tokenization from word to WP causes inaccurate labels to be generated. This paper proposes and studies two confidence models of increasing complexity to solve this problem. The final model uses self-attention to directly learn word-level confidence without needing subword tokenization, and exploits full context features from multiple hypotheses to improve confidence accuracy. Experiments on Voice Search and long-tail test sets show standard metrics (e.g., NCE, AUC, RMSE) improving substantially. The proposed confidence module also enables a model selection approach to combine an on-device E2E model with a hybrid model on the server to address the rare word recognition problem for the E2E model.

eess.AS↗

arXiv2021

Transformer Based Deliberation for Two-Pass Speech Recognition

Interactive speech recognition systems must generate words quickly while also producing accurate results. Two-pass models excel at these requirements by employing a first-pass decoder that quickly emits words, and a second-pass decoder that requires more context but is more accurate. Previous work has established that a deliberation network can be an effective second-pass model. The model attends to two kinds of inputs at once: encoded audio frames and the hypothesis text from the first-pass model. In this work, we explore using transformer layers instead of long-short term memory (LSTM) layers for deliberation rescoring. In transformer layers, we generalize the "encoder-decoder" attention to attend to both encoded audio and first-pass text hypotheses. The output context vectors are then combined by a merger layer. Compared to LSTM-based deliberation, our best transformer deliberation achieves 7% relative word error rate improvements along with a 38% reduction in computation. We also compare against non-deliberation transformer rescoring, and find a 9% relative improvement.

cs.CL↗

arXiv2020

Realtime CNN-based Keypoint Detector with Sobel Filter and CNN-based Descriptor Trained with Keypoint Candidates

The local feature detector and descriptor are essential in many computer vision tasks, such as SLAM and 3D reconstruction. In this paper, we introduce two separate CNNs, lightweight SobelNet and DesNet, to detect key points and to compute dense local descriptors. The detector and the descriptor work in parallel. Sobel filter provides the edge structure of the input images as the input of CNN. The locations of key points will be obtained after exerting the non-maximum suppression (NMS) process on the output map of the CNN. We design Gaussian loss for the training process of SobelNet to detect corner points as keypoints. At the same time, the input of DesNet is the original grayscale image, and circle loss is used to train DesNet. Besides, output maps of SobelNet are needed while training DesNet. We have evaluated our method on several benchmarks including HPatches benchmark, ETH benchmark, and FM-Bench. SobelNet achieves better or comparable performance with less computation compared with SOTA methods in recent years. The inference time of an image of 640x480 is 7.59ms and 1.09ms for SobelNet and DesNet respectively on RTX 2070 SUPER.

cs.CV↗