Filters








29 Hits in 1.6 sec

Learning Domain-Invariant Relationship with Instrumental Variable for Domain Generalization [article]

Junkun Yuan, Xu Ma, Kun Kuang, Ruoxuan Xiong, Mingming Gong, Lanfen Lin
2021 arXiv   pre-print
Domain generalization (DG) aims to learn from multiple source domains a model that generalizes well on unseen target domains. Existing methods mainly learn input feature representations with invariant marginal distribution, while the invariance of the conditional distribution is more essential for unknown domain generalization. This paper proposes an instrumental variable-based approach to learn the domain-invariant relationship between input features and labels contained in the conditional
more » ... ribution. Interestingly, with a causal view on the data generating process, we find that the input features of one domain are valid instrumental variables for other domains. Inspired by this finding, we design a simple yet effective framework to learn the Domain-invariant Relationship with Instrumental VariablE (DRIVE) via a two-stage IV method. Specifically, it first learns the conditional distribution of input features of one domain given input features of another domain, and then it estimates the domain-invariant relationship by predicting labels with the learned conditional distribution. Simulation experiments show the proposed method accurately captures the domain-invariant relationship. Extensive experiments on several datasets consistently demonstrate that DRIVE yields state-of-the-art results.
arXiv:2110.01438v1 fatcat:uwvoxpo66jb7hccl7er4dalvp4

Data-Driven Adaptive Simultaneous Machine Translation [article]

Guangxu Xun, Mingbo Ma, Yuchen Bian, Xingyu Cai, Jiaji Huang, Renjie Zheng, Junkun Chen, Jiahong Yuan, Kenneth Church, Liang Huang
2022 arXiv   pre-print
In simultaneous translation (SimulMT), the most widely used strategy is the wait-k policy thanks to its simplicity and effectiveness in balancing translation quality and latency. However, wait-k suffers from two major limitations: (a) it is a fixed policy that can not adaptively adjust latency given context, and (b) its training is much slower than full-sentence translation. To alleviate these issues, we propose a novel and efficient training scheme for adaptive SimulMT by augmenting the
more » ... g corpus with adaptive prefix-to-prefix pairs, while the training complexity remains the same as that of training full-sentence translation models. Experiments on two language pairs show that our method outperforms all strong baselines in terms of translation quality and latency.
arXiv:2204.12672v1 fatcat:olomv4g3zbdufkfha5zgqcyuaq

Learning Decomposed Representation for Counterfactual Inference [article]

Anpeng Wu, Kun Kuang, Junkun Yuan, Bo Li, Runze Wu, Qiang Zhu, Yueting Zhuang, Fei Wu
2021 arXiv   pre-print
The fundamental problem in treatment effect estimation from observational data is confounder identification and balancing. Most of the previous methods realized confounder balancing by treating all observed pre-treatment variables as confounders, ignoring further identifying confounders and non-confounders. In general, not all the observed pre-treatment variables are confounders that refer to the common causes of the treatment and the outcome, some variables only contribute to the treatment and
more » ... some only contribute to the outcome. Balancing those non-confounders, including instrumental variables and adjustment variables, would generate additional bias for treatment effect estimation. By modeling the different causal relations among observed pre-treatment variables, treatment and outcome, we propose a synergistic learning framework to 1) identify confounders by learning decomposed representations of both confounders and non-confounders, 2) balance confounder with sample re-weighting technique, and simultaneously 3) estimate the treatment effect in observational studies via counterfactual inference. Empirical results on synthetic and real-world datasets demonstrate that the proposed method can precisely decompose confounders and achieve a more precise estimation of treatment effect than baselines.
arXiv:2006.07040v2 fatcat:mw4twlenybevxgc6bflzzsqj5y

Domain-Specific Bias Filtering for Single Labeled Domain Generalization [article]

Junkun Yuan, Xu Ma, Defang Chen, Kun Kuang, Fei Wu, Lanfen Lin
2021 arXiv   pre-print
Domain generalization (DG) utilizes multiple labeled source datasets to train a generalizable model for unseen target domains. However, due to expensive annotation costs, the requirements of labeling all the source data are hard to be met in real-world applications. In this paper, we investigate a Single Labeled Domain Generalization (SLDG) task with only one source domain being labeled, which is more practical and challenging than the Conventional Domain Generalization (CDG). A major obstacle
more » ... n the SLDG task is the discriminability-generalization bias: discriminative information in the labeled source dataset may contain domain-specific bias, constraining the generalization of the trained model. To tackle this challenging task, we propose a novel method called Domain-Specific Bias Filtering (DSBF), which initializes a discriminative model with the labeled source data and then filters out its domain-specific bias with the unlabeled source data for generalization improvement. We divide the filtering process into (1) feature extractor debiasing via k-means clustering-based semantic feature re-extraction and (2) classifier calibrating through attention-guided semantic feature projection. DSBF unifies the exploration of the labeled and the unlabeled source data to enhance the discriminability and generalization of the trained model, resulting in a highly generalizable model. We further provide theoretical analysis to verify the proposed domain-specific bias filtering process. Extensive experiments on multiple datasets show the superior performance of DSBF in tackling both the challenging SLDG task and the CDG task.
arXiv:2110.00726v2 fatcat:p7sxwp2tkjgtlecjyy7zluoiaa

Auto IV: Counterfactual Prediction via Automatic Instrumental Variable Decomposition [article]

Junkun Yuan, Anpeng Wu, Kun Kuang, Bo Li, Runze Wu, Fei Wu, Lanfen Lin
2021 arXiv   pre-print
Instrumental variables (IVs), sources of treatment randomization that are conditionally independent of the outcome, play an important role in causal inference with unobserved confounders. However, the existing IV-based counterfactual prediction methods need well-predefined IVs, while it's an art rather than science to find valid IVs in many real-world scenes. Moreover, the predefined hand-made IVs could be weak or erroneous by violating the conditions of valid IVs. These thorny facts hinder the
more » ... application of the IV-based counterfactual prediction methods. In this paper, we propose a novel Automatic Instrumental Variable decomposition (AutoIV) algorithm to automatically generate representations serving the role of IVs from observed variables (IV candidates). Specifically, we let the learned IV representations satisfy the relevance condition with the treatment and exclusion condition with the outcome via mutual information maximization and minimization constraints, respectively. We also learn confounder representations by encouraging them to be relevant to both the treatment and the outcome. The IV and confounder representations compete for the information with their constraints in an adversarial game, which allows us to get valid IV representations for IV-based counterfactual prediction. Extensive experiments demonstrate that our method generates valid IV representations for accurate IV-based counterfactual prediction.
arXiv:2107.05884v1 fatcat:27yoxdo7pfav3kyqat5xdcvpim

PaddleSpeech: An Easy-to-Use All-in-One Speech Toolkit [article]

Hui Zhang, Tian Yuan, Junkun Chen, Xintong Li, Renjie Zheng, Yuxin Huang, Xiaojie Chen, Enlei Gong, Zeyu Chen, Xiaoguang Hu, Dianhai Yu, Yanjun Ma (+1 others)
2022 arXiv   pre-print
PaddleSpeech is an open-source all-in-one speech toolkit. It aims at facilitating the development and research of speech processing technologies by providing an easy-to-use command-line interface and a simple code structure. This paper describes the design philosophy and core architecture of PaddleSpeech to support several essential speech-to-text and text-to-speech tasks. PaddleSpeech achieves competitive or state-of-the-art performance on various speech datasets and implements the most
more » ... methods. It also provides recipes and pretrained models to quickly reproduce the experimental results in this paper. PaddleSpeech is publicly avaiable at https://github.com/PaddlePaddle/PaddleSpeech.
arXiv:2205.12007v1 fatcat:zfubo5stfvczhaaeepxod4u6hy

TorchDrug: A Powerful and Flexible Machine Learning Platform for Drug Discovery [article]

Zhaocheng Zhu, Chence Shi, Zuobai Zhang, Shengchao Liu, Minghao Xu, Xinyu Yuan, Yangtian Zhang, Junkun Chen, Huiyu Cai, Jiarui Lu, Chang Ma, Runcheng Liu (+3 others)
2022 arXiv   pre-print
Machine learning has huge potential to revolutionize the field of drug discovery and is attracting increasing attention in recent years. However, lacking domain knowledge (e.g., which tasks to work on), standard benchmarks and data preprocessing pipelines are the main obstacles for machine learning researchers to work in this domain. To facilitate the progress of machine learning for drug discovery, we develop TorchDrug, a powerful and flexible machine learning platform for drug discovery built
more » ... on top of PyTorch. TorchDrug benchmarks a variety of important tasks in drug discovery, including molecular property prediction, pretrained molecular representations, de novo molecular design and optimization, retrosynthsis prediction, and biomedical knowledge graph reasoning. State-of-the-art techniques based on geometric deep learning (or graph machine learning), deep generative models, reinforcement learning and knowledge graph reasoning are implemented for these tasks. TorchDrug features a hierarchical interface that facilitates customization from both novices and experts in this domain. Tutorials, benchmark results and documentation are available at https://torchdrug.ai. Code is released under Apache License 2.0.
arXiv:2202.08320v1 fatcat:jzbsxbhizjgwflwvh2avoyzcai

Prognostic and predictive value of FCER1G in glioma outcomes and response to immunotherapy

Houshi Xu, Qingwei Zhu, Lan Tang, Junkun Jiang, Huiwen Yuan, Anke Zhang, Meiqing Lou
2021 Cancer Cell International  
Purpose Glioma is the most prevalent malignant form of brain tumors, with a dismal prognosis. Currently, cancer immunotherapy has emerged as a revolutionary treatment for patients with advanced highly aggressive therapy-resistant tumors. However, there is no effective biomarker to reflect the response to immunotherapy in glioma patient so far. So we aim to assess the clinical predictive value of FCER1G in patients with glioma. Methods The expression level and correlation between clinical
more » ... is and FER1G levels were analyzed with the data from CGGA, TCGA, and GEO database. Univariate and multivariate cox regression model was built to predict the prognosis of glioma patients with multiple factors. Then the correlation between FCER1G with immune cell infiltration and activation was analyzed. At last, we predict the immunotherapeutic response in both high and low FCER1G expression subgroups. Results FCER1G was significantly higher in glioma with greater malignancy and predicted poor prognosis. In multivariate analysis, the hazard ratio of FCER1G expression (Low versus High) was 0.66 and 95 % CI is 0.54 to 0.79 (P < 0.001), whereas age (HR = 1.26, 95 % CI 1.04–1.52), grade (HR = 2.75, 95 % CI 2.06–3.68), tumor recurrence (HR = 2.17, 95 % CI 1.81–2.62), IDH mutant (HR = 2.46, 95 % CI 1.97–3.01) and chemotherapeutic status (HR = 1.4, 95 % CI 1.20–1.80) are also included. Furthermore, we illustrated that gene FCER1G stratified glioma cases into high and low FCER1G expression subgroups that demonstrated with distinct clinical outcomes and T cell activation. At last, we demonstrated that high FCER1G levels presented great immunotherapeutic response in glioma patients. Conclusions This study demonstrated FCER1G as a novel predictor for clinical diagnosis, prognosis, and response to immunotherapy in glioma patient. Assess expression of FCER1G is a promising method to discover patients that may benefit from immunotherapy.
doi:10.1186/s12935-021-01804-3 pmid:33579299 fatcat:un5uonh5qvhrfhmo7jpwdbxdiy

Auto IV: Counterfactual Prediction via Automatic Instrumental Variable Decomposition

Junkun Yuan, Anpeng Wu, Kun Kuang, Bo Li, Runze Wu, Fei Wu, Lanfen Lin
2022 ACM Transactions on Knowledge Discovery from Data  
Instrumental variables (IVs), sources of treatment randomization that are conditionally independent of the outcome, play an important role in causal inference with unobserved confounders. However, the existing IV-based counterfactual prediction methods need well-predefined IVs, while it's an art rather than science to find valid IVs in many real-world scenes. Moreover, the predefined hand-made IVs could be weak or erroneous by violating the conditions of valid IVs. These thorny facts hinder the
more » ... application of the IV-based counterfactual prediction methods. In this article, we propose a novel Automatic Instrumental Variable decomposition (AutoIV) algorithm to automatically generate representations serving the role of IVs from observed variables (IV candidates). Specifically, we let the learned IV representations satisfy the relevance condition with the treatment and exclusion condition with the outcome via mutual information maximization and minimization constraints, respectively. We also learn confounder representations by encouraging them to be relevant to both the treatment and the outcome. The IV and confounder representations compete for the information with their constraints in an adversarial game, which allows us to get valid IV representations for IV-based counterfactual prediction. Extensive experiments demonstrate that our method generates valid IV representations for accurate IV-based counterfactual prediction.
doi:10.1145/3494568 fatcat:5nwm36iu6rfphku3og47s5tdym

Synthesis and Characterization of Block Copolymers of Poly(silylene diethynylbenzen) and Poly(silylene dipropargyl aryl ether)

Man Gao, Chengyuan Shang, Jixian Li, Gang Han, Junkun Tang, Qiaolong Yuan, Farong Huang
2021 Polymers  
Poly(silylene diethynylbenzene)–b–poly(silylene dipropargyloxy diphenyl propane) copolymer (ABA-A), poly(silylene diethynylbenzene)–b–poly(silylene dipropargyloxy diphenyl ether) copolymer (ABA-O), and a contrast poly(silylene diethynylbenzene) with equivalent polymerization degree were synthesized through Grignard reactions. The structures and properties of the copolymers were investigated via hydrogen nuclear magnetic resonance, Fourier transform infrared spectroscopy, Haake torque rheometer,
more » ... differential scanning calorimetry, dynamic mechanical analysis, thermogravimetric analysis and mechanical tests. The results show that the block copolymers possess comprehensive properties, especially good processability and good mechanical properties. The processing windows of these copolymers are wider than 58 °C. The flexural strength of the cured ABA-A copolymer reaches as high as 40.2 MPa. The degradation temperatures at 5% weight loss (Td5) of the cured copolymers in nitrogen are all above 560 °C.
doi:10.3390/polym13091511 pmid:34067206 fatcat:mycof552hjawrjpxeocbpnz3ye

Subgraph Networks with Application to Structural Feature Space Expansion [article]

Qi Xuan, Jinhuan Wang, Minghao Zhao, Junkun Yuan, Chenbo Fu, Zhongyuan Ruan, Guanrong Chen
2019 arXiv   pre-print
Yuan, and C.  ... 
arXiv:1903.09022v2 fatcat:ko6wdvy325dwjk4xkck4vct6vy

Collaborative Semantic Aggregation and Calibration for Separated Domain Generalization [article]

Junkun Yuan, Xu Ma, Defang Chen, Kun Kuang, Fei Wu, Lanfen Lin
2022 arXiv   pre-print
(Junkun Yuan and Xu Ma contributed equally to this work.) J. Yuan, X. Ma, D. Chen, K. Kuang, F. Wu, and L.  ... 
arXiv:2110.06736v3 fatcat:tmmfsgritjcjbjg62pp4vw35oi

VATEX: A Large-Scale, High-Quality Multilingual Dataset for Video-and-Language Research [article]

Xin Wang, Jiawei Wu, Junkun Chen, Lei Li, Yuan-Fang Wang, William Yang Wang
2020 arXiv   pre-print
We present a new large-scale multilingual video description dataset, VATEX, which contains over 41,250 videos and 825,000 captions in both English and Chinese. Among the captions, there are over 206,000 English-Chinese parallel translation pairs. Compared to the widely-used MSR-VTT dataset, VATEX is multilingual, larger, linguistically complex, and more diverse in terms of both video and natural language descriptions. We also introduce two tasks for video-and-language research based on VATEX:
more » ... ) Multilingual Video Captioning, aimed at describing a video in various languages with a compact unified captioning model, and (2) Video-guided Machine Translation, to translate a source language description into the target language using the video information as additional spatiotemporal context. Extensive experiments on the VATEX dataset show that, first, the unified multilingual model can not only produce both English and Chinese descriptions for a video more efficiently, but also offer improved performance over the monolingual models. Furthermore, we demonstrate that the spatiotemporal video context can be effectively utilized to align source and target languages and thus assist machine translation. In the end, we discuss the potentials of using VATEX for other video-and-language research.
arXiv:1904.03493v3 fatcat:mo5gwbvjhza5thhftq6ivyehci

Identification of Environmental Factors Associated with Inflammatory Bowel Disease in a Southwestern Highland Region of China: A Nested Case-Control Study

Junkun Niu, Jiarong Miao, Yuan Tang, Qiong Nan, Yan Liu, Gang Yang, Xiangqian Dong, Qi Huang, Shuxian Xia, Kunhua Wang, Yinglei Miao, John Green
2016 PLoS ONE  
The aim of this study was to examine environmental factors associated with inflammatory bowel disease (IBD) in Yunnan Province, a southwestern highland region of China. Methods In this nested case-control study, newly diagnosed ulcerative colitis (UC) cases in 2 cities in Yunnan Province and Crohn's disease (CD) cases in 16 cities in Yunnan Province were recruited between 2008 and 2013. Controls were matched by geography, sex and age at a ratio of 1:4. Data were collected using the designed
more » ... tionnaire. Conditional logistic regression models were used to estimate adjusted odds ratios (ORs). Results A total of 678 UC and 102 CD cases were recruited. For UC, various factors were associated with an increased risk of developing UC: dietary habits, including frequent irregular meal times; consumption of fried foods, salty foods and frozen dinners; childhood factors, including intestinal infectious diseases and frequent use of antibiotics; and other factors, such as mental labor, high work stress, use of non-aspirin non-steroidal anti-inflammatory drugs and allergies (OR > 1, p < 0.05). Other factors showed a protective effect: such as consumption of fruits, current smoking, physical activity, and drinking tea (OR < 1, p < 0.05). For CD, appendectomy and irregular meal times increased the disease risk (OR >1, p < 0.05), whereas physical activity may have reduced this risk (OR < 1, p < 0.05).
doi:10.1371/journal.pone.0153524 pmid:27070313 pmcid:PMC4829194 fatcat:uildpy5rijbyrb55avpynkg4ay

Large-scale multiferroic complex oxide epitaxy with magnetically switched polarization enabled by solution processing

Cong Liu, Feng An, Paria S M Gharavi, Qinwen Lu, Junkun Zha, Chao Chen, Liming Wang, Xiaozhi Zhan, Zedong Xu, Yuan Zhang, Ke Qu, Junxiang Yao (+14 others)
2019 National Science Review  
Complex oxides with tunable structures have many fascinating properties, though high-quality complex oxide epitaxy with precisely controlled composition is still out of reach. Here we have successfully developed solution-based single crystalline epitaxy for multiferroic (1-x)BiTi(1-y)/2FeyMg(1-y)/2O3-(x)CaTiO3 (BTFM-CTO) solid solution in large area, confirming its ferroelectricity at atomic-scale with a strong spontaneous polarization. Careful compositional tuning leads to a bulk magnetization
more » ... of 0.07 ± 0.035 μB/Fe at room temperature, enabling magnetically induced polarization switching exhibiting a large magnetoelectric coefficient of 2.7–3.0 × 10−7 s/m. This work demonstrates the great potential of solution processing in large-scale complex oxide epitaxy and establishes novel room-temperature magnetoelectric coupling in epitaxial BTFM-CTO film, making it possible to explore a much wider space of composition, phase, and structure that can be easily scaled up for industrial applications.
doi:10.1093/nsr/nwz143 pmid:34692020 pmcid:PMC8289034 fatcat:rwpq64hi4vbwlpjyy7unkvcaci
« Previous Showing results 1 — 15 out of 29 results