全校師生：
我校定于2020年04月22日舉辦研究生靈犀學(xué)術(shù)殿堂——夏志明教授報(bào)告會(huì)，現(xiàn)將有關(guān)事項(xiàng)通知如下：
1.報(bào)告會(huì)簡(jiǎn)介
報(bào)告人：夏志明教授
時(shí)間：2020年04月22日（星期三）下午3：00（開(kāi)始時(shí)間）
地點(diǎn)：騰訊會(huì)議，ID：561530041
報(bào)告題目：Deep PCA: A methodology of feature extraction and dimension reduction for high-order data
內(nèi)容簡(jiǎn)介：Facing with rapidly-increasing demands for analyzing high-order data or multiway data, feature-extracting methods become imperative for analysis and processing. The traditional feature-extracting methods, however, either need to overly vectorize the data and smash the original structure hidden in data, such as PCA and PCA-like methods, which is unfavourable to the data recovery, or can not eliminate the redundant information very well, such as Tucker Decomposition (TD) and TD-like methods. To overcome these limitations, we propose a more flexible and more powerful tool, called the Deep Principal Components Analysis (Deep-PCA) in this paper. By segmenting a random tensor into equal-sized subarrays named \textit{sections} and maximizing variations caused by orthogonal projections of these \textit{sections}, the Deep-PCA finds principal components in a parsimonious and flexible way. In so doing, two new operations on tensors, the $S$-\textit{direction inner/outer product}, are introduced to formulate tensor projection and recovery. With different segmentation ways characterized by \textit{section depth} and \textit{direction}, the Deep-PCA can be implemented many times in different ways, which defines the sequential and global Deep-PCA respectively. These multiple Deep-PCA take the PCA and PCA-like, Tucker Decomposition and the TD-like as the special cases, which corresponds to the deepest section-depth and the shallowest section depth respectively. We propose an adaptive depth and direction selection algorithm for implementation of Deep-PCA. The Deep-PCA is then tested in terms of subspace recovery ability, compression ability and feature extraction performance when applied to a set of artificial data, surveillance videos and hyperspectral imaging data. All the tests support the flexibility, effectiveness and usefulness of Deep-PCA.
2.歡迎各學(xué)院師生前來(lái)聽(tīng)報(bào)告。報(bào)告會(huì)期間請(qǐng)關(guān)閉手機(jī)或?qū)⑹謾C(jī)調(diào)至靜音模式。
黨委學(xué)生工作部
數(shù)學(xué)與統(tǒng)計(jì)學(xué)院
2020年4月20日
報(bào)告人簡(jiǎn)介
西北大學(xué)數(shù)學(xué)學(xué)院教授，博士生導(dǎo)師，西北大學(xué)現(xiàn)代統(tǒng)計(jì)研究中心副主任，主要致力于張量數(shù)據(jù)分析、大數(shù)據(jù)異質(zhì)性結(jié)構(gòu)推斷、分布式統(tǒng)計(jì)推斷與計(jì)算、生物統(tǒng)計(jì)學(xué)等數(shù)據(jù)科學(xué)理論與應(yīng)用研究。在“Biometrika”、“Journal of machine learning research”,“Technometrics”、“Statistics in Medicine”、“Journal of Statistical Planning and Inference”、“Statistics”等國(guó)際統(tǒng)計(jì)與機(jī)器學(xué)習(xí)期刊以及“中國(guó)科學(xué)”、“應(yīng)用概率統(tǒng)計(jì)”等國(guó)內(nèi)期刊發(fā)表論文30余篇；主持國(guó)家自然科學(xué)基金項(xiàng)目3項(xiàng)，主持省部級(jí)項(xiàng)目3項(xiàng),作為骨干成員獲得“陜西省科學(xué)技術(shù)進(jìn)步獎(jiǎng)”二、三等獎(jiǎng)共2項(xiàng)，“陜西省高?？茖W(xué)技術(shù)獎(jiǎng)”一等獎(jiǎng)共2項(xiàng)，“陜西省國(guó)防科技進(jìn)步獎(jiǎng)”一等獎(jiǎng)1項(xiàng)；先后赴香港科技大學(xué)、佛羅里達(dá)大學(xué)等科研機(jī)構(gòu)進(jìn)行專(zhuān)業(yè)訪(fǎng)問(wèn)與學(xué)術(shù)交流。
