讲座人简介：

韩晓旭教授是纽约福特汉姆大学计算机与信息科学系终身教授，也是哥伦比亚大学蛋白质组学中心附属教研员。1992年，韩晓旭教授在bv伟德国际体育获得数学学士学位。2001年与2004年，韩晓旭教授在爱荷华大学数学与计算科学系分别获得了计算机科学硕士学位和应用博士学位，在爱荷华老员工物学比较基因组学中心从事生物信息学博士后研究。2010年，他晋升为副教授，获授终身教职。2012秋季，韩晓旭教授加入福特汉姆大学。他已在国际著名期刊发表了近50篇论文，内容涉及到生物信息学，大数据，数据挖掘，机器学习和数据分析领域。同时，他也担任计算机与信息科学系林肯中心的副主席。

讲座题目1：Data-driven Massive Data Analytics in Modern Data Science

讲座日期：2019-06-11

讲座时间：15:00

讲座内容简介：

The surge of all kind of massive data pushes scientific research to move from model-driven to data-driven. The key question is how to do such a move? This talk covers recently completed or ongoing data-driven analytics projects in modern data science. It mainly includes endowment analytics for U.S. wealthiest universities, implied volatility analytics for options in finance, AI in real estate analytics. We demonstrate essential components of data-driven analytics from data collection & preprocessing, modeling construction & selection, and knowledge discovery in the projects, besides presenting high-resolution analytical solutions to state-of-the-art data science challenges.  It is noted that the endowment analytics and real-estate analytics are subfields in data science firstly initialized in our work. Furthermore, the integrative learning techniques presented in implied volatility analytics can be extended as a generic machine learning approach in massive data analytics.

讲座题目2：Big data analytics for precision medicine and high-frequency trading

讲座日期：2019-06-12

讲座时间：15:00

讲座内容简介：

All kinds of big data present unique challenges in different fields.  This talk mainly handles big omics data in precision medicine and big finance data from high-frequency trading. We present novel algorithms and techniques to dig knowledge from big omics data that has millions of variables and present it in a meaningful approach for the sake of disease diagnosis. Our studies in big omics data solve urgent problems in big data and precision medicine. The first is how to visualize big data via targeted mining; The second is how to conduct a high-performance early complex disease discovery via big data, besides presenting a novel solution for high-dimensional imbalanced big data.
Alternatively, we first present as a section-volatility via a rigorous big data analytics approach to estimate volatility for high-frequency trading data. Compared to state-of-the-art realized-volatility estimation, the proposed section volatility captures the behavior of high-frequency trading more accurately.

讲座题目3：High frequency trading market marker prediction and price discovery via manifold learning and deep learning

讲座日期：2019-06-27

讲座时间：15:00

讲座内容简介：

High-frequency trading (HFT) has dominated the financial market in recent years. It counts about 55% of trading volumes in the U.S. equity market and takes almost 80% of foreign exchange futures volumes. As automated trading relying on computer algorithms to conduct trading in a high-frequency mode, it can finish a transaction in a few milliseconds. Unlike traditional trading, it can make a profit from a marginal or even tiny price change by trading large volumes of securities in seconds.
The rise of high-frequency trading brings challenges both in finance and data science for its ultra-fast trading speed and huge data size. In this work, we propose a novel trading marker prediction approach in HFT by employing manifold learning: locally linear embedding (LLE) and state-of-the-art clustering techniques: Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The proposed method demonstrates its advantage over its peers in trading marker prediction. Furthermore, we present effective deep learning models in price discovery and profitable high-frequency trading and compare its performance with those of peers. To the best of our knowledge, it is the first work in Fintech to address market marker prediction and HFT trading modeling problems.