Jian Tian

Zhejiang University

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Xi Li

Nanchang Institute of Science & Technology

Tao Wei

China National Petroleum Corporation (China)

Jin Di Wang

Hubei University of Technology

Ning An

Bengbu Medical College

Bin Li

Alibaba Group (China)

Yong Niu

Chinese Academy of Sciences

Fengqi Yu

Shenzhen Institutes of Advanced Technology

Shi Feng Shi

Hanbin Zhao

Zhejiang University

Fei Wu

Zhejiang University

Cooperative Institutions

Zhejiang University

Zhejiang University of Science and Technology

Chinese Academy of Sciences

University of Chinese Academy of Sciences

Hubei University of Technology

Nanchang University

Northwestern Polytechnical University

Yunnan University

State Grid Corporation of China (China)

University of Science and Technology of China

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Multitask Non-Autoregressive Model for Human Motion Prediction

IEEE Transactions on Image Processing (2020)

Bin Li Jian Tian Zhongfei Zhang Hailin Feng Xi Li

Human motion prediction, which aims at predicting future human skeletons given the past ones, is a typical sequence-to-sequence problem. Therefore, extensive efforts have been devoted to exploring different RNN-based encoder-decoder architectures. However, by generating target poses conditioned on the previously generated ones, these models are prone to bringing issues such as error accumulation problem. In this paper, we argue that such issue is mainly caused by adopting autoregressive manner. Hence, a novel Non-AuToregressive model (NAT) is proposed with a complete non-autoregressive decoding scheme, as well as a context encoder and a positional encoding module. More specifically, the context encoder embeds the given poses from temporal and spatial perspectives. The frame decoder is responsible for predicting each future pose independently. The positional encoding module injects positional signal into the model to indicate the temporal order. Besides, a multitask training paradigm is presented for both low-level human skeleton prediction and high-level human action recognition, resulting in the considerable improvement for the prediction task. Our approach is evaluated on Human3.6M and CMU-Mocap benchmarks and outperforms state-of-the-art autoregressive methods.

ENCODE

Sequence (biology)

10.1109/tip.2020.3038362

Cite

Citations (40)

The Effect of Remelting on the Microstructure and Mechanical Properties of a Nickel Superalloy

Materials science forum (2016)

Zhi Wei Zhang Yong Niu Jian Tian Ning An Yang Gao

Effect of remelting processes on the microstructure and mechanical properties of as-cast Nimonic 90 superalloy was investigated by OM (optical micrograph), SEM (scanning electron microscopy), EDS (energy-dispersive X-ray) and tensile tests. The results indicated that the microstructure of the as-cast alloy was mainly composed of γ, γ' and carbides which contain Ti and Cr elements. The average grain size of the alloy tends to increase with mould shell temperature ranging from 350°C to 950°C. The strength of the as-cast alloy decreased with the increasing of mold shell temperature, with constant elongation as the mold shell temperature changes.

Nimonic

Elongation

10.4028/www.scientific.net/msf.849.492

Cite

Citations (0)

Unsupervised Domain Adaptation for Image Classification via Structure-Conditioned Adversarial Learning

arXiv (Cornell University) (2021)

Hui Wang Jian Tian Songyuan Li Hanbin Zhao Qi Tian

Unsupervised domain adaptation (UDA) typically carries out knowledge transfer from a label-rich source domain to an unlabeled target domain by adversarial learning. In principle, existing UDA approaches mainly focus on the global distribution alignment between domains while ignoring the intrinsic local distribution properties. Motivated by this observation, we propose an end-to-end structure-conditioned adversarial learning scheme (SCAL) that is able to preserve the intra-class compactness during domain distribution alignment. By using local structures as structure-aware conditions, the proposed scheme is implemented in a structure-conditioned adversarial learning pipeline. The above learning procedure is iteratively performed by alternating between local structures establishment and structure-conditioned adversarial learning. Experimental results demonstrate the effectiveness of the proposed scheme in UDA scenarios.

Domain Adaptation

Contextual image classification

10.48550/arxiv.2103.02808

Cite

Citations (5)

Solidification behavior, microstructure and silicon twinning of Al-10Si alloys with ytterbium addition

Journal of Rare Earths (2018)

Zhi Hu Zhou Dong Zheng Yin Hong Yan Jian Tian

Recalescence

10.1016/j.jre.2017.12.007

Cite

Citations (19)

Joint Event Extraction Based on Global Event-Type Guidance and Attention Enhancement

Computers, materials & continua/Computers, materials & continua (Print) (2021)

Daojian Zeng Jian Tian Ruoyao Peng Jianhua Dai Hui Gao

10.32604/cmc.2021.017028

Cite

Citations (1)

Effect of Step Quenching and Zr Addition on the Microstructure and Mechanical Properties of Al-Mg-Si Alloy

Korean Journal of Metals and Materials (2014)

Jian Tian Kee Do Woo Kwang Jin Lee Yuyong Chen

The effects of substituting a step quenching process for the traditional water quenching and of adding zirconium into an Al-Mg-Si alloy, were investigated using differential scanning calorimetry (DSC) and transmission electron microscope (TEM), together with a tensile and hardness tester. The results indicated that step quenching and zirconium addition affected the microstructure and mechanical properties significantly. Step quenching at 100 ℃ for the Al-Mg-Si alloy was the optimal choice because it promoted the precipitation of the β′′ phase, afforded the maximum age-hardening response and strength increment, and restrained the negative effect of natural aging. Zirconium addition improved the strength while retaining good plasticity for the precipitation of Al3Zr. Zr partially promoted the precipitation kinetics of the β′′ phase when step quenching was conducted at 50 or 100 ℃. †(Received January 15, 2013)

10.3365/kjmm.2014.52.2.101

Cite

Citations (12)

Parameter influence on speech recognition rate of modified RBF neural network

International Conference on Intelligent Control and Information Processing (2010)

Xia Wang Jian Tian Mengjun Wang

In RBF neural network designing, hidden neuron number and parameter influence the performance of network. The paper discusses influences of pruning hidden neurons using different criteria and parameter on speech recognition rate of modified RBF neural network. First we introduce three hidden neuron pruning criteria, then propose a modified RBF neural network, at last recognition results before and after pruning hidden neuron, relationship among constant C of BIC, hidden neuron number and recognition rates are given. Experiments show that proper criteria and parameter will conduce to better results.

Pruning

10.1109/icicip.2010.5565218

Cite

Citations (0)

Noisy speech recognition based on modified RBF network

Xia Wang Jian Tian Xiaoqun Zhao

Speech recognition rate is often influenced by noise because of mismatching between training and recognition environments. This paper present a recognition system based on modified RBF Network. In this system, training data are predistorted by morphology filter same as recognizing data. For modified RBF network, cluster centers are decided by competitive learning algorithm, weights are achieved using conjugate gradient descent algorithm, network structure is optimized with pruning hidden neurons, the parameters of the hidden layer are trained dynamically. Experiments show that this system can increase system performance in noisy environment.

Pruning

10.1109/cinc.2010.5643738

Cite

Citations (0)

Studies on Geology of the Backfilling Fly Ash and the Foundation Treatments

Advanced materials research (2011)

Jian Tian Bao Shan Xia Jin Di Wang Tao Wei

Whether the site of backfilling fly ash was appropriate for engineering geology, and how to use effective means to change the geologic bearing capacity of the site were urgent solving problems. This work took a site of backfilling fly ash in the Beishiwang Village, Yindu District, Anyang City, Henan Province as object of study, detailedly did a complete geologic survey on it, got related data and made its engineering geologic assessment and according to the characteristics and requirements of construction projects, we select the most optimal treatment means of the foundation and adopt the most safe and economic perfusion pile with post-pressure grouting to strengthen this engineering foundation.

Foundation (evidence)

Research Object

Geological survey

10.4028/www.scientific.net/amr.243-249.3182

Cite

Citations (1)

Improving Search by Extending Tags According to Recommendation Level and Combinations of Types

Jian Tian Kening Gao Yin Zhang Bin Zhang

As the collaborative tagging systems such as Delicious, Flickr and Last. fm become more and more popular, a large amount of resources produced by publishers, together with rich semantic metadata, become available. There are lots of ways to make use of tag information and so far the most discussed usage is for searching. The distribution of tag types differs greatly across different systems. Also the distribution shows large difference between publishers and searchers. In order to expand tags of resources for publishers and keywords for searchers reasonable, this paper shows a comparison of the distributions of both kinds of users and proposes an approach which could calculate the recommendation level of types. The level of types could be used to describe whether the type is valuable to the corresponding users. The distribution of different combination of types has also been investigated, and with such information we analysed the most popular combination of types in query log across different collections. We compare the searching accuracy on original datasets against the datasets with resources after being expanded tags by our proposed methods. Experimental results show that our method could improve searching accuracy.

10.1109/skg.2011.17

Cite

Citations (3)