Search

Results

Data Science Seminar: 90 Hits

dept/SMLSem/index.html|even earlier talks]]. {{ :seminars:sml:sml_app.png?direct&700 |}} ==== Fall 2024 ==== ... e Response Regression with Hidden Variables.\\ [[seminars:datasci:101524|Abstract]] * **October 22, 20... nding auction with unobserved heterogeneity.\\ [[seminars:datasci:102224|Abstract]] * **November 5, 20... g robust loss functions in machine learning.\\ [[seminars:datasci:110524|Abstract]] * **November 12, 20

Previous talks in the Data Science Seminar: 76 Hits

ent Dimension Reduction in High Dimension**\\ [[:seminars:sml:160927|Abstract]] * **October 18**\\ //Speake... ommunity detection with nodal information**\\ [[:seminars:sml:161018|Abstract]] * **October 25**\\ //Speake... ting Graphical Structure among Predictors**\\ [[:seminars:sml:161025|Abstract]] * **November 1**\\ //Speake... ling Smart Urban Surveillance at The Edge**\\ [[:seminars:sml:161101|Abstract]] * **November 8**\\ //Speake

Machine Learning Working Group: 18 Hits

* Topic: //MM Principle in Optimization// * {{seminars:sml:mlwg:Notes01242017.pdf|Note}} taken by: **Lin Yao... ngle-based Multicategory Classification// * {{seminars:sml:mlwg:Notes02142017.pdf|Note}} taken by: **Yuan Fa... sant: **Chen Liang** * Topic: //TBD// * {{seminars:sml:mlwg:Notes02212017.pdf|Note}} taken by: **TBD** ... ssant: **Yuan Fang** * Topic: //TBD// * {{seminars:sml:mlwg:Notes02282017.pdf|Note}} taken by: **TBD** =

April 25, 2017: 2 Hits

<WRAP centeralign>##Data Science Seminar##\\ Hosted by Department of Mathematical Sciences</WRAP> ~~META:title=April 25, 2017~~ * Date: Tuesday, April 25, 2017 * Time: 12:00-1:00 * Room: WH-100E * Speaker: Lingzhou Xue (Pennsylvania State University) * Title: Sufficient Forecasting Using Factor Models <WRAP center box 80%> <WRAP centeralign>**//Abstract//**</WRAP> We consider forecasting a single time series using a large number of predictors when a nonlinear forecasting function is present. The linear forecasting is very appealing due to its simplicity. However, it only reveals one dimension of the predictive power in the underlying factors. This paper develops a new forecasting methodology called the sufficient forecasting, which provides several sufficient predictive indices to deliver additional predictive power. The sufficient forecasting correctly estimates projections of the underlying factors even in the presence of an arbitrary and unknown forecasting function. Our work identifies the effective factors that have impacts on the forecast target when the target and the cross-sectional predictors are driven by different sets of common factors. We derive asymptotic properties for the estimate of the central subspace spanned by these projection directions as well as the estimates of the sufficient predictive indices. We also prove that when the assumed linear forecasting function is violated, the simple linear estimate actually falls into this central subspace. Our method and theory allow the number of predictors to be larger than the number of observations. We finally demonstrate that the sufficient forecasting improves upon the linear forecasting in both simulation studies and an empirical study of forecasting macroeconomic variables. </WRAP> More details about the Data Science seminar can be found at https://www2.math.binghamton.edu/p/seminars/sml

April 18, 2017: 2 Hits

<WRAP centeralign>##Data Science Seminar##\\ Hosted by Department of Mathematical Sciences</WRAP> ~~META:title=April 18, 2017~~ * Date: Tuesday, April 18, 2017 * Time: 12:00-1:00 * Room: WH-100E * Speaker: Changqing Cheng (SSIE at Binghamton University) * Title: Heterogeneous Recurrence Monitoring of Dynamic Transients in Ultraprecision Machining Processes <WRAP center box 80%> <WRAP centeralign>**//Abstract//**</WRAP> In-situ monitoring and control of process variations are important for quality assurance in ultraprecision machining (UPM) processes. Recent advance in sensing and communication technology have fueled increasing interests to develop sensor-based monitoring approaches for anomaly detection in the UPM process. However, conventional approaches are limited in their ability to address the complex dynamics hidden in the nonlinear and nonstationary processes. As a result, it is difficult for them to effectively capture the process variations of UPM. We present a new heterogeneous recurrence monitoring approach for in-situ monitoring and predictive control of the UPM process, and it has the capability to detect the shift from stable to unstable cutting in UPM processes with an average run length of 1. </WRAP> More details about the Data Science seminar can be found at https://www2.math.binghamton.edu/p/seminars/sml

April 20, 2017: 2 Hits

<WRAP centeralign>##Data Science Seminar##\\ Hosted by Department of Mathematical Sciences</WRAP> ~~META:title=April 20, 2017~~ * Date: Thursday, April 20, 2017 (note special day and time) * Time: 1:15--2:15 * Room: WH-100E * Speaker: Qiusheng Wu (Geography at Binghamton University) * Title: Environmental Monitoring and Analysis with Geospatial Big Data Powered by Google <WRAP center box 80%> <WRAP centeralign>**//Abstract//**</WRAP> Earth observation data have been collected by various satellite and airborne sensors through remote sensing technologies since the 1970s. On the one hand, the petabyte archives of remote sensing imagery are crucial for many environmental and societal applications such as surface water mapping, deforestation tracking, drought monitoring, disaster response, disease surveillance and so on. On the other hand, the ever-increasing volume and variety of geospatial big data pose great challenges in data storage, management, analysis, and visualization. Until recently, global-scale environmental analyses with fine-resolution remote sensing imagery have been very limited due to the excessive computational cost associated with these geospatial big data. Google Earth Engine is a new cloud computing platform that combines a multi-petabyte catalog of satellite imagery and geospatial datasets with planetary-scale analysis capabilities, allowing scientists and researchers to access and analyze vast amounts of satellite imagery free of charge. The platform offers parallel computational access to thousands of computers in Google's data centers. This talk will demonstrate the application of Google Earth Engine and high-resolution light detection and ranging (LiDAR) data for various environmental applications, such as wetland mapping and monitoring. </WRAP> More details about the Data Science seminar can be found at https://www2.math.binghamton.edu/p/seminars/sml

May 9, 2017: 2 Hits

<WRAP centeralign>##Data Science Seminar##\\ Hosted by Department of Mathematical Sciences</WRAP> ~~META:title=May 9, 2017~~ * Date: Tuesday, May 9, 2017 * Time: 12:00-1:00 * Room: WH-100E * Speaker: Yuying Xie (Michigan State University) * Title: Joint Estimation of Multiple Dependent Gaussian Graphical Models with Applications to Mouse Genomics <WRAP center box 80%> <WRAP centeralign>**//Abstract//**</WRAP> Gaussian graphical models are widely used to represent conditional dependence among random variables. In this paper, we propose a novel estimator for data arising from a group of Gaussian graphical models that are themselves dependent. A motivating example is that of modeling gene expression collected on multiple tissues from the same individual: here the multivariate outcome is affected by dependencies acting not only at the level of the specific tissues, but also at the level of the whole body; existing methods that assume independence among graphs are not applicable in this case. To estimate multiple dependent graphs, we decompose the problem into two graphical layers: the systemic layer, which affects all outcomes and thereby induces cross-graph dependence, and the category-specific layer, which represents graph-specific variation. We propose a graphical EM technique that estimates both layers jointly, establish estimation consistency and selection sparsistency of the proposed estimator, and confirm by simulation that the EM method is superior to a simple one-step method. We apply our technique to mouse genomics data and obtain biologically plausible results. </WRAP> More details about the Data Science seminar can be found at https://www2.math.binghamton.edu/p/seminars/sml