arXiv subjects

Yike Zhang

Publications and source records attributed to Yike Zhang.

49 records · Page 3Linked to original sources

arXiv2023

Using two point correlation functions to understand the assembly histories of Milky Way-like galaxies

The two point correlation function (2PCF) is a powerful statistical tool to measure galaxy clustering. Although 2PCF has also been used to study the clustering of stars on parsec and sub-parsec scales, its physical implication is not clear on such non-linear scales. In this study, we use the Illustris-TNG50 simulation to study the connection between the 2PCF signals of accreted halo stars and the assembly histories of Milky Way-mass galaxies. We find, in general, the 2PCF signal increases with the increase in galactocentric radii, $r$, and with the decrease in the pair separations. Galaxies which assemble late on average have stronger 2PCF signals. With $z_{1/4}$, $z_{1/2}$ and $z_{3/4}$ defined as the redshifts when galaxies accreted one-fourth, half and three-fourths of their ex-situ stellar mass at today, we find all of them show the strongest correlations with the 2PCF signals at $r\sim0.2R_{200}$. $z_{3/4}$ shows the strongest correlations at all radii than those of $z_{1/4}$ or $z_{1/2}$, as later accreted stars preserve better clusterings. However, the correlations between the 2PCF signals at different radii and the galaxy formation times all have large scatters. The 2PCFs in velocity space show weaker correlations with the galaxy formation times within $0.38R_{200}$ than real space 2PCFs, and the scatter is considerably large. Both the real and velocity space 2PCFs correlate with the assembly histories of the host dark matter halos as well. Within $0.38R_{200}$, the real space 2PCF shows stronger correlations with the galaxy formation histories than with the halo formation histories, while the velocity space 2PCFs do not show large differences. We conclude that it is difficult to use 2PCF alone to precisely predict the formation times or assembly histories of galaxies.

astro-ph.GA↗

arXiv2023

Two Stage Contextual Word Filtering for Context bias in Unified Streaming and Non-streaming Transducer

It is difficult for an E2E ASR system to recognize words such as entities appearing infrequently in the training data. A widely used method to mitigate this issue is feeding contextual information into the acoustic model. Previous works have proven that a compact and accurate contextual list can boost the performance significantly. In this paper, we propose an efficient approach to obtain a high quality contextual list for a unified streaming/non-streaming based E2E model. Specifically, we make use of the phone-level streaming output to first filter the predefined contextual word list then fuse it into non-casual encoder and decoder to generate the final recognition results. Our approach improve the accuracy of the contextual ASR system and speed up the inference process. Experiments on two datasets demonstrates over 20% CER reduction comparing to the baseline system. Meanwhile, the RTF of our system can be stabilized within 0.15 when the size of the contextual word list grows over 6,000.

cs.SD↗

arXiv2023

Self-supervised Registration and Segmentation of the Ossicles with A Single Ground Truth Label

AI-assisted surgeries have drawn the attention of the medical image research community due to their real-world impact on improving surgery success rates. For image-guided surgeries, such as Cochlear Implants (CIs), accurate object segmentation can provide useful information for surgeons before an operation. Recently published image segmentation methods that leverage machine learning usually rely on a large number of manually predefined ground truth labels. However, it is a laborious and time-consuming task to prepare the dataset. This paper presents a novel technique using a self-supervised 3D-UNet that produces a dense deformation field between an atlas and a target image that can be used for atlas-based segmentation of the ossicles. Our results show that our method outperforms traditional image segmentation methods and generates a more accurate boundary around the ossicles based on Dice similarity coefficient and point-to-point error comparison. The mean Dice coefficient is improved by 8.51% with our proposed method.

eess.IV↗

arXiv2023

Vector-based Efficient Data Hiding in Encrypted Images via Multi-MSB Replacement

As an essential technique for data privacy protection, reversible data hiding in encrypted images (RDHEI) methods have drawn intensive research interest in recent years. In response to the increasing demand for protecting data privacy, novel methods that perform RDHEI are continually being developed. We propose two effective multi-MSB (most significant bit) replacement-based approaches that yield comparably high data embedding capacity, improve overall processing speed, and enhance reconstructed images' quality. Our first method, Efficient Multi-MSB Replacement-RDHEI (EMR-RDHEI), obtains higher data embedding rates (DERs, also known as payloads) and better visual quality in reconstructed images when compared with many other state-of-the-art methods. Our second method, Lossless Multi-MSB Replacement-RDHEI (LMR-RDHEI), can losslessly recover original images after an information embedding process is performed. To verify the accuracy of our methods, we compared them with other recent RDHEI techniques and performed extensive experiments using the widely accepted BOWS-2 dataset. Our experimental results showed that the DER of our EMR-RDHEI method ranged from 1.2087 bit per pixel (bpp) to 6.2682 bpp with an average of 3.2457 bpp. For the LMR-RDHEI method, the average DER was 2.5325 bpp, with a range between 0.2129 bpp and 6.0168 bpp. Our results demonstrate that these methods outperform many other state-of-the-art RDHEI algorithms. Additionally, the multi-MSB replacement-based approach provides a clean design and efficient vectorized implementation.

cs.CR↗

arXiv2022

Leveraging Acoustic Contextual Representation by Audio-textual Cross-modal Learning for Conversational ASR

Leveraging context information is an intuitive idea to improve performance on conversational automatic speech recognition(ASR). Previous works usually adopt recognized hypotheses of historical utterances as preceding context, which may bias the current recognized hypothesis due to the inevitable historicalrecognition errors. To avoid this problem, we propose an audio-textual cross-modal representation extractor to learn contextual representations directly from preceding speech. Specifically, it consists of two modal-related encoders, extracting high-level latent features from speech and the corresponding text, and a cross-modal encoder, which aims to learn the correlation between speech and text. We randomly mask some input tokens and input sequences of each modality. Then a token-missing or modal-missing prediction with a modal-level CTC loss on the cross-modal encoder is performed. Thus, the model captures not only the bi-directional context dependencies in a specific modality but also relationships between different modalities. Then, during the training of the conversational ASR system, the extractor will be frozen to extract the textual representation of preceding speech, while such representation is used as context fed to the ASR decoder through attention mechanism. The effectiveness of the proposed approach is validated on several Mandarin conversation corpora and the highest character error rate (CER) reduction up to 16% is achieved on the MagicData dataset.

eess.AS↗

arXiv2022

Censer: Curriculum Semi-supervised Learning for Speech Recognition Based on Self-supervised Pre-training

Recent studies have shown that the benefits provided by self-supervised pre-training and self-training (pseudo-labeling) are complementary. Semi-supervised fine-tuning strategies under the pre-training framework, however, remain insufficiently studied. Besides, modern semi-supervised speech recognition algorithms either treat unlabeled data indiscriminately or filter out noisy samples with a confidence threshold. The dissimilarities among different unlabeled data are often ignored. In this paper, we propose Censer, a semi-supervised speech recognition algorithm based on self-supervised pre-training to maximize the utilization of unlabeled data. The pre-training stage of Censer adopts wav2vec2.0 and the fine-tuning stage employs an improved semi-supervised learning algorithm from slimIPL, which leverages unlabeled data progressively according to their pseudo labels' qualities. We also incorporate a temporal pseudo label pool and an exponential moving average to control the pseudo labels' update frequency and to avoid model divergence. Experimental results on Libri-Light and LibriSpeech datasets manifest our proposed method achieves better performance compared to existing approaches while being more unified.

cs.SD↗

arXiv2022

A practical framework for multi-domain speech recognition and an instance sampling method to neural language modeling

Automatic speech recognition (ASR) systems used on smart phones or vehicles are usually required to process speech queries from very different domains. In such situations, a vanilla ASR system usually fails to perform well on every domain. This paper proposes a multi-domain ASR framework for Tencent Map, a navigation app used on smart phones and in-vehicle infotainment systems. The proposed framework consists of three core parts: a basic ASR module to generate n-best lists of a speech query, a text classification module to determine which domain the speech query belongs to, and a reranking module to rescore n-best lists using domain-specific language models. In addition, an instance sampling based method to training neural network language models (NNLMs) is proposed to address the data imbalance problem in multi-domain ASR. In experiments, the proposed framework was evaluated on navigation domain and music domain, since navigating and playing music are two main features of Tencent Map. Compared to a general ASR system, the proposed framework achieves a relative 13% $\sim$ 22% character error rate reduction on several test sets collected from Tencent Map and our in-car voice assistant.

eess.AS↗

arXiv2022

Improving CTC-based speech recognition via knowledge transferring from pre-trained language models

Recently, end-to-end automatic speech recognition models based on connectionist temporal classification (CTC) have achieved impressive results, especially when fine-tuned from wav2vec2.0 models. Due to the conditional independence assumption, CTC-based models are always weaker than attention-based encoder-decoder models and require the assistance of external language models (LMs). To solve this issue, we propose two knowledge transferring methods that leverage pre-trained LMs, such as BERT and GPT2, to improve CTC-based models. The first method is based on representation learning, in which the CTC-based models use the representation produced by BERT as an auxiliary learning target. The second method is based on joint classification learning, which combines GPT2 for text modeling with a hybrid CTC/attention architecture. Experiment on AISHELL-1 corpus yields a character error rate (CER) of 4.2% on the test set. When compared to the vanilla CTC-based models fine-tuned from the wav2vec2.0 models, our knowledge transferring method reduces CER by 16.1% relatively without external LMs.

cs.CL↗

arXiv2022

Conversational Speech Recognition By Learning Conversation-level Characteristics

Conversational automatic speech recognition (ASR) is a task to recognize conversational speech including multiple speakers. Unlike sentence-level ASR, conversational ASR can naturally take advantages from specific characteristics of conversation, such as role preference and topical coherence. This paper proposes a conversational ASR model which explicitly learns conversation-level characteristics under the prevalent end-to-end neural framework. The highlights of the proposed model are twofold. First, a latent variational module (LVM) is attached to a conformer-based encoder-decoder ASR backbone to learn role preference and topical coherence. Second, a topic model is specifically adopted to bias the outputs of the decoder to words in the predicted topics. Experiments on two Mandarin conversational ASR tasks show that the proposed model achieves a maximum 12% relative character error rate (CER) reduction.

cs.SD↗

arXiv2021

Improving Hybrid CTC/Attention End-to-end Speech Recognition with Pretrained Acoustic and Language Model

Recently, self-supervised pretraining has achieved impressive results in end-to-end (E2E) automatic speech recognition (ASR). However, the dominant sequence-to-sequence (S2S) E2E model is still hard to fully utilize the self-supervised pre-training methods because its decoder is conditioned on acoustic representation thus cannot be pretrained separately. In this paper, we propose a pretrained Transformer (Preformer) S2S ASR architecture based on hybrid CTC/attention E2E models to fully utilize the pretrained acoustic models (AMs) and language models (LMs). In our framework, the encoder is initialized with a pretrained AM (wav2vec2.0). The Preformer leverages CTC as an auxiliary task during training and inference. Furthermore, we design a one-cross decoder (OCD), which relaxes the dependence on acoustic representations so that it can be initialized with pretrained LM (DistilGPT2). Experiments are conducted on the AISHELL-1 corpus and achieve a $4.6\%$ character error rate (CER) on the test set. Compared with our vanilla hybrid CTC/attention Transformer baseline, our proposed CTC/attention-based Preformer yields $27\%$ relative CER reduction. To the best of our knowledge, this is the first work to utilize both pretrained AM and LM in a S2S ASR system.

eess.AS↗

arXiv2021

Improving Streaming Transformer Based ASR Under a Framework of Self-supervised Learning

Recently self-supervised learning has emerged as an effective approach to improve the performance of automatic speech recognition (ASR). Under such a framework, the neural network is usually pre-trained with massive unlabeled data and then fine-tuned with limited labeled data. However, the non-streaming architecture like bidirectional transformer is usually adopted by the neural network to achieve competitive results, which can not be used in streaming scenarios. In this paper, we mainly focus on improving the performance of streaming transformer under the self-supervised learning framework. Specifically, we propose a novel two-stage training method during fine-tuning, which combines knowledge distilling and self-training. The proposed training method achieves 16.3% relative word error rate (WER) reduction on Librispeech noisy test set. Finally, by only using the 100h clean subset of Librispeech as the labeled data and the rest (860h) as the unlabeled data, our streaming transformer based model obtains competitive WERs 3.5/8.7 on Librispeech clean/noisy test sets.

eess.AS↗

arXiv2021

Improving Speech Recognition Accuracy of Local POI Using Geographical Models

Nowadays voice search for points of interest (POI) is becoming increasingly popular. However, speech recognition for local POI has remained to be a challenge due to multi-dialect and massive POI. This paper improves speech recognition accuracy for local POI from two aspects. Firstly, a geographic acoustic model (Geo-AM) is proposed. The Geo-AM deals with multi-dialect problem using dialect-specific input feature and dialect-specific top layer. Secondly, a group of geo-specific language models (Geo-LMs) are integrated into our speech recognition system to improve recognition accuracy of long tail and homophone POI. During decoding, specific language models are selected on demand according to users' geographic location. Experiments show that the proposed Geo-AM achieves 6.5%$\sim$10.1% relative character error rate (CER) reduction on an accent testset and the proposed Geo-AM and Geo-LM totally achieve over 18.7% relative CER reduction on Tencent Map task.

eess.AS↗

arXiv2020

A Template-Based Approach to the Photometric Classification of SN 1991bg-like Supernovae in the SDSS-II Supernova Survey

The use of Type Ia Supernovae (SNe Ia) to measure cosmological parameters has grown significantly over the past two decades. However, there exists a significant diversity in the SN Ia population that is not well understood. Over-luminous SN 1991T-like and sub-luminous SN 1991bg-like objects are two characteristic examples of peculiar SNe. The identification and classification of such objects is an important step in studying what makes them unique from the remaining SN population. With the upcoming Vera C. Rubin Observatory promising on the order of a million new SNe over a ten-year survey, spectroscopic classifications will be possible for only a small subset of observed targets. As such, photometric classification has become an increasingly important concern in preparing for the next generation of astronomical surveys. Using observations from the Sloan Digital Sky Survey II (SDSS-II) SN Survey, we apply here an empirically based classification technique targeted at the identification of SN 1991bg-like SNe in photometric data sets. By performing dedicated fits to photometric data in the rest-frame redder and bluer bandpasses, we classify 16 previously unidentified 91bg-like SNe. Using SDSS-II host-galaxy measurements, we find that these SNe are preferentially found in host galaxies having an older average stellar age than the hosts of normal SNe Ia. We also find that these SNe are found at a further physical distance from the center of their host galaxies. We find no statistically significant bias in host galaxy mass or specific star formation rate for these targets.

astro-ph.IM↗