徐亦达老师一直致力于国内的机器学习推广,在他授权下,本站公布徐亦达老师的精选论文70篇,以及他上课的课件,并提供下载
徐亦达,现任悉尼科技大学教授,UTS全球数据技术中心机器学习和数据分析实验室主任。主要研究方向是机器学习,数据分析和计算机视觉。他在国际重要期刊与会议发表数篇高影响因子论文;编写了大量的数理统计、概率和机器学习教材。
精选论文
徐亦达老师和他的团队精选了70篇论文,其中包括了非参贝叶斯算法的研究和应用,行列式点过程,3D计算机视觉,带约束下几何优化,非负矩阵分解,视频跟踪,GAN的文字到图像产生等课题 。
原文的github:https://github.com/roboticcam/publications
文末提供下载
精选期刊论文
- Li, Y., Li, K., Xu, R. Y. D., Wang, X., (2020), Exploring Temporal Consistency for Human Pose Estimation in Videos, accepted Jan 2020, Pattern Recognition, IF 5.898
- Li, C., Xie, H., Fan, X, Xu, R. Y. D., Van Huffel, S., Mengersen K., (2020), Kernelized Sparse Bayesian Matrix Factorization, accepted Feb 2020 IEEE Transactions on Neural Networks and Learning Systems, IF 11.683
- Li M., Xu, R. Y. D., Xin, J., Zhang, K., Jing, J., (2020) Fast non-rigid points registration with cluster correspondences projection, 170, 107425, Signal Processing, IF 4.086
- Li, C., Xie, H., Fan, Xu, R.Y.D., Van Huffel, S., Mengersen K., (2019), Image denoising based on nonlocal Bayesian singular value thresholding and Stein's unbiased risk estimator, 28 (10), pp. 4899 – 4911 IEEE Transactions on Image Processing, IF 6.79
- Jiang S., Li K, Xu, R.Y.D., (2019) Relative Pairwise Relationship Constrained Non-negative Matrix Factorisation, Vol. 31, No. 8, pp.1595 – 1609 IEEE Transactions on Knowledge and Data Engineering, IF 3.857
- Bargi, A., Xu, R.Y.D., & Piccardi, M., (2018) AdOn HDP-HMM: An Adaptive Online Model for Segmentation and Classification of Sequential Data, 29 (9), pp. 3953 - 3968 IEEE Transactions on Neural Networks and Learning Systems IF 11.683
- Xuan, J., Lu, J., Zhang, G., Xu, R.Y.D., & Luo, X., (2017) Doubly Nonparametric Sparse Nonnegative Matrix Factorization based on Dependent Indian Buffet Processes, Vol 29, NO. 5 pp. 1835 - 1849 IEEE Transactions on Neural Networks and Learning Systems IF 11.683
- Li, J., Deng, C., Xu, R.Y.D., Tao, D., & Zhao, B., (2017), Robust Object Tracking with Discrete Graph Based Multiple Experts, Vol 26, Issue 6, pp. 2736 - 2750, IEEE Transactions On Image Processing, IF 6.79
- Xuan, J., Lu, J., Zhang, G., Xu, R.Y.D., Luo, X., (2017) Bayesian Nonparametric Relational Topic Model through Dependent Gamma Processes, Vol.29, Issue 7, pp. 1357 – 1369 IEEE Transactions on Knowledge and Data Engineering IF 3.857
- Fan, X., Xu, R.Y.D., Cao, L., Song. Y., (2017), Learning Nonparametric Relational Models by Conjugately Incorporating Node Information in a Network. Vol 47(3), pp. 589 – 599 IEEE Transaction on Cybernetics, IF 10.387
- Xuan, J., Lu, J., Zhang, G., Xu, R.Y.D., Luo, X., (2017), A Bayesian Nonparametric Model for Multi-Label Learning, Volume 106, Issue 11, pp 1787–1815 Machine Learning IF 2.809
- Li, J., Zhao, B., Deng, C., & Xu, R.Y.D., (2016), Time Varying Metric Learning for visual tracking, vol. 80, pp. 157-164. Pattern Recognition Letters, IF 2.810
- Qiao, M., Xu, R.Y.D., Bian, W. & Tao, D. (2016), Fast sampling for time-varying Determinantal Point Processes, vol. 11, no. 1. ACM Transactions on Knowledge Discovery from Data, IF 1.000
- Kemp, M., Xu, R.Y.D., (2015), Geometrically-constrained balloon fitting for multiple connected ellipses, vol. 48, no. 7, pp. 2198-2208. Pattern Recognition, IF 5.898
- Qiao, M., Bian, W., Xu, R.Y.D., Tao, D., (2015), Diversified Hidden Markov Models for Sequential Labeling, vol. 27, no. 11, pp. 2947-2960. IEEE Transactions on Knowledge and Data Engineering, IF 3.857
- Fan, X., Cao, L., Xu, R.Y.D., (2015), Dynamic Infinite Mixed-Membership Stochastic Blockmodel, Vol. 26, Issue 9, pp. 2072 - 2085, IEEE Transactions on Neural Networks and Learning Systems IF 11.683
- Zare Borzeshi, E., Concha, O.P., Xu, R.Y.D., & Piccardi, M. (2013), Joint Action Segmentation and Classification by an Extended Hidden Markov Model, vol. 20, no. 12, pp. 1207-1210. IEEE Signal Processing Letters, IF 3.268
- Xu, R.Y.D., & Kemp, M. (2010), Fitting Multiple Connected Ellipses To An Image Silhouette Hierarchically, vol. 19, no. 7, pp. 1673-1682. IEEE Transactions On Image Processing, IF 6.79
- Xu, R.Y.D., & Kemp, M. (2010), An Iterative Approach for Fitting Multiple Connected Ellipse Structure to Silhouette, vol. 31, no. 13, pp. 1860-1867. Pattern Recognition Letters, IF 2.810
arXiv论文
- Xu, R.Y.D., Caron, F., Doucet., A (2016), Bayesian nonparametric image segmentation using a generalized Swendsen-Wang algorithm, arXiv:1602.03048
精选会议论文
- Huang, W., Xu, R. Y. D., Du, W., Zeng Y., and Zhao Y., (2020) Mean field theory for deep dropout networks: digging up gradient backpropagation deeply (to appear), the 24th European Conference on Artificial Intelligence arXiv version(ECAI 2020)
- Li, Y., Li., K, Jiang., S, Zhang., Z Y, Huang., C Z T and Xu, R. Y. D., (2020) Geometry Self-Supervised method for 3D Human Pose, Association for the Advancement of Artificial Intelligence (AAAI 2020)
- Huang, W., Xu, R.Y.D., Oppermann, I., (2019), Realistic Image Generation using Region-phrase Attention, Asian Conference eon Machine Learning, PMLR 101:284-299, 2019, (ACML 2019)
- Huang, W., Xu, R.Y.D., Oppermann, I., (2019), Efficient Diversified Mini-Batch Selection using Variable High-layer Features, Asian Conference on Machine Learning, PMLR 101:300-315, 2019, (ACML 2019)
- Fan., X, Xu., R. Y. D., Cao., L (2016), Copula Mixed-Membership Stochastic Blockmodel, International Joint Conference on Artificial Intelligence (IJCAI 2016)
- Li, Q, Bian., W, Xu., R. Y. D., You., J and Tao., D (2016), Random Mixed Field Model for Mixed-Attribute Data Restoration, Association for the Advancement of Artificial Intelligence (AAAI 2016)
- Xuan, J., Lu, J., Zhang, G., Xu, R.Y.D., Luo, X.,(2015), Infinite Author Topic Model based on Mixed Gamma-Negative Binomial Process, IEEE International Conference on Data Mining (ICDM 2015)
- Li, M., Xu, R. Y.D., & He, X.J. 2015, Face hallucination based on Nonparametric Bayesian learning, IEEE International Conference on Image Processing, (ICIP 2015) Quebec City, Canada, pp. 986-990.
- Bargi, A., Xu, R. Y. D., Ghahramani Z, Piccardi., M (2014), A non-parametric conditional factor regression model for high-dimensional input and response, Seventeenth International Conference on Artificial Intelligence and Statistics (AISTAT 2014), pp.77-85
- Bargi A., Xu, R. Y. D., Piccardi M, (2014), An Infinite Adaptive Online Learning Model for Segmentation and Classification of Streaming Data, International Conference on Pattern Recognition (ICPR 2014)
- Xu, R. Y. D., Kemp, M., (2009), Multiple Curvature Based Approach to Human Upper Body Parts Detection with Connected Ellipse Model Fine-Tuning, IEEE International Conference on Image Processing (ICIP2009), Cairo Egypt
- Xu, R. Y. D., Brown, J., Traish, J., Dezwa, D., (2008), A Computer Vision Based Camera Pedestal's Vertical Motion Control. International Conference on Pattern Recognition (ICPR 2008), Florida, USA
- Xu, R. Y. D., (2008). A Computer Vision based Whiteboard Capture System, IEEE Workshop on Application of Computer Vision (WACV 2008), Colorado, USA
- Xu, R. Y. D., Gao, J., Antolovich, M., (2008), Novel Methods for High-Resolution Facial Image Capture using Calibrated PTZ and Static Cameras, IEEE International Conference on Multimedia & Expo (ICME 2008), Hanover, Germany: 45-48.
- Allen, J. G., Xu, R. Y. D., Jin, J., (2003), Object Tracking Using CamShift Algorithm and Multiple Quantized Feature Spaces, Pan-Sydney Area Workshop on Visual Information Processing (VIP2003), Sydney, Australia: 3-7 480 Google citations
课件目录
徐亦达老师课件的github:
https://github.com/roboticcam/machine-learning-notes
前言
- 笔记的视频演示
- 最近的研究讲义
- 噪声对比估计 (Noise Contrastive Estimation), 概率密度再参数化
一、基础知识
1.概率论与数理统计基础
- 贝叶斯模型
- 概率估计
- 统计属性
2.概率模型
- E-M算法
- 状态空间模型(动态模型)
3.高级概率模型
- 贝叶斯非参数(BNP)及其推导基础
- 贝叶斯非参数(BNP)扩展
- 行列式点过程
4.推导课件
- 变分推导
- 随机矩阵
- 蒙特卡洛简介
- 马尔可夫链蒙特卡洛
- 粒子滤波器(序列蒙特卡洛)
二、深度学习
- 优化方法
- 神经网络
- 卷积神经网络:从基础到最近的研究
- 词表示和近似Softmax
- 深度自然语言处理
- 深度增强学习
- 受限玻尔兹曼机
三、数据科学
- 30分钟介绍人工智能和机器学习
- 回归方法
- 推荐系统
- 降维
- 数据分析简介和相关的jupyternotebook
四、致谢
徐亦达老师的课件的内容简介及相关文件
前言
- 笔记的视频演示
简介:
徐亦达老师在2015年用中文录制了这些课件中约20%的内容 (备注:课件为全英文),大家可以在YouTube、哔哩哔哩 以及优酷观看和下载。
YouTube:
https://www.youtube.com/channel/UConITmGn5PFr0hxTI2tWD4Q
哔哩哔哩:
https://space.bilibili.com/327617676
优酷:
http://i.youku.com/i/UMzIzNDgxNTg5Ng
- 最近的研究讲义
2018年7月在创新工厂和北大合办的DeeCamp的课件(当概率遇到神经网络)
简介:
主题包括:EM算法和矩阵胶囊网络;行列式点过程和神经网络压缩; 卡尔曼滤波器和LSTM; 模型估计和二分类问题关系。
文件:DeeCamp2018_Xu_final.pptx
- 噪声对比估计 (Noise Contrastive Estimation), 概率密度再参数化
简介:
噪声对比估计 (Noise Contrastive Estimation), 概率密度再参数化以及自然梯度的详细说明。
文件:selected_probability.pdf
一、基础知识
1.概率论与数理统计基础
- 贝叶斯模型
简介:
贝叶斯模型的修订包括贝叶斯预测模型,条件期望。
文件:bayesian.pdf
- 概率估计
简介:一些有用的分布,共轭,MLE,MAP,指数族和自然参数。
文件:probability.pdf
- 统计属性
简介:有用的统计属性可以帮助我们证明事物,包括切比雪夫和马尔科夫不等式。
文件:statistics.pdf
2.概率模型
- E-M算法
简介:E-M的收敛证明,E-M到高斯混合模型的例子。
文件:em.pdf,gmm_demo.m,kmeans_demo.m
优酷链接:http://v.youku.com/v_show/id_XMTM1MjY1MDU5Mg
- 状态空间模型(动态模型)
简介:
详细解释了卡尔曼滤波器和隐马尔可夫模型。
文件:
dynamic_model.pdf,kalman_demo.m
优酷链接:
隐马尔可夫模型:http://v.youku.com/v_show/id_XMTM1MzQ1NDk5Ng
卡尔曼滤波:http://v.youku.com/v_show/id_XMTM2ODU1MzMzMg
3.高级概率模型
- 贝叶斯非参数(BNP)及其推导基础
简介:
Dircihlet Process(DP),中国餐厅流程见解。
文件:
non_parametrics.pdf ,dirichlet_process.m, chinese_restaurant_process.ipynb
优酷链接:
http://v.youku.com/v_show/id_XMTM3NDY0MDkxNg
- 贝叶斯非参数(BNP)扩展
简介:
分层DP,HDP-HMM,印度自助餐过程(IBP)。
文件:non_parametrics_extensions.pdf
- 行列式点过程
简介:
解释DPP的边际分布,L-ensemble,其抽样策略,我们在时变DPP中的工作细节。
文件:dpp.pdf
4.推导课件
- 变分推导
简介:
解释变分贝叶斯非指数和指数族分布加上随机变分推导。
文件:
variational.pdf ,vbnormalgamma.m
优酷链接:
http://v.youku.com/v_show/id_XMTM1Njc5NzkxNg
- 随机矩阵
简介:
随机矩阵,幂方法收敛定理,详细平衡和PageRank算法。
文件:stochastic_matrices.pdf
- 蒙特卡洛简介
简介:
逆CDF,消除,自适应消除,重要性采样。
文件:
introduction_monte_carlo.pdf ,adaptiverejectionsampling.m,hybrid_gmm.m
- 马尔可夫链蒙特卡洛
简介:
M-H, Gibbs, SliceSampling,Elliptical Slice sampling, Swendesen-Wang, demonstrate collapsed GibbsusingLDA
文件:
markov_chain_monte_carlo.pdf,ldagibbsexample.m ,test_autocorrelation.m, gibbs.m
优酷链接:
http://v.youku.com/v_show/id_XMTM1NjAyNDYyNA
- 粒子滤波器(序列蒙特卡洛)
简介:
连续蒙特卡罗方法,冷凝滤波算法,辅助粒子滤波器。
文件:particle_filter.pdf
优酷链接:
http://v.youku.com/v_show/id_XMTM3MTE1Mjk2OA
二、深度学习
- 优化方法
简介:
常用的优化方法。不仅限于深度学习。
文件:optimization.pdf
- 神经网络
简介:
基本神经网络和多层感知器。
文件:neural_networks.pdf
- 卷积神经网络:从基础到最近的研究
简介:
卷积神经网络的详细解释,各种损失函数,中心损失函数,对比损失函数,残差网络,YOLO,SSD。
文件:cnn_beyond.pdf
- 词表示和近似Softmax
简介:
Word2Vec, skip-gram, GloVe, 噪声对比估计,负采样,Gumbel-max技巧。
文件:
word_vector.pdf
- 深度自然语言处理
简介:
RNN,LSTM,具有注意力机制的Seq2Seq,集束搜索,RNN,LSTM,具有注意力机制的Seq2Seq,集束搜索,Attention is all you need,卷积Seq2Seq,指针网络。
文件:deep_nlp.pdf
- 深度增强学习
简介:
强化学习的基础知识,马尔可夫决策过程,贝尔曼方程,并进入深度Q学习(正在建设中)
文件:dqn.pdf
- 受限玻尔兹曼机
简介:
受限玻尔兹曼机(RBM)的基础知识
文件:rbm_gan.pdf
三、数据科学
- 30分钟介绍人工智能和机器学习
简介:
用30分钟来介绍人工智能和机器学习。 (感谢徐亦达老师的博士生常皓东进行协助编辑)
文件:30_min_AI.pptx
- 回归方法
简介:
分类:Logistic回归和Softmax分类;回归:线性回归,多项式回归,混合效果模型。
文件:regression.pdf, costFunction.m,soft_max.m
- 推荐系统
简介:
协同过滤,分解机,非负矩阵分解,乘法更新规则。
文件:recommendation.pdf
- 降维
简介:
典型的PCA和 t-SNE。
文件:dimension_reduction.pdf
- 数据分析简介和相关的jupyternotebook
简介:
机器学习和数据科学的三个视角。 监督与无监督学习,分类准确性。
文件:AI_and_machine_learning.pdf
下载
可以在徐亦达老师的github直接下载:
https://github.com/roboticcam
由于github速度慢,推荐用百度云下载:
https://pan.baidu.com/s/1Z30LzdspxQ4h0IAGOdMyew 提取码: s9b9
若被和谐请回复“徐亦达”获取下载地址。