“News-20140526”版本间的差异

来自cslt Wiki
跳转至: 导航搜索
Lyy讨论 | 贡献
 
(2位用户的4个中间修订版本未显示)
第1行: 第1行:
 +
IBM scientists Yong Qin、Songfang Huang、Ming Li visit CSLT
  
== IBM中国科学家秦勇、黄松访、李敏访问语音和语言技术中心 ==
+
Time: 2014-05-26 <br/>
 +
Location: ROOM 1-305, BLDG FIT, Tsinghua University
  
5月26日下午,来自IBM中国研究实验室(IBM China Research Lab, CRL) 的三位高级科学家访问我院语音中心并就IBM在多项前沿研究中的最新进展做了精彩报告。 秦勇博士主要介绍了IBM在认知计算方面的研究计划和研究现状。 秦博士的报告从宏观上展示了IBM在认知计算方面的总体部署、实现途径和现有积累,提出在大数据时代,基于认知计算对结构化、半结构化和非结构化数据进行处理和挖掘,实现人与计算机的共生环境,最终为人们提供智能化的人机交互方法和推理能力,为企业解决复杂问题提供决策支持。 黄松芳博士主要介绍了语音识别和机器学习领域中语言模型技术的最新进展,总结了传统n-gram模型的平滑方法,介绍了贝叶斯模型、最大熵模型以及神经网络模型等高级语言的建模技术。 李敏博士主要介绍了移动支付背景下IBM的移动生物认证的解决方案和技术进展,特别介绍了基于视频和人脸识别的认证方法以及将声音、图象、笔迹等多模态认证方法相融合的技术,展望了生物认证技术在移动互联网和移动金融等方面的应用前景。
+
Cognitive Computing
 +
 
 +
•      Presenter:
 +
–      Kelvin Qin/秦勇
 +
–      Senior Technical Staff Member
 +
–      IBM Research – China
 +
•      Abstract:
 +
In the era of Big Data, the biggest challenge for enterprises today is to re-discover the trends, re-discover the clients, re-define the relationship with customers to satisfy their ever growing demands. We strive to improve enterprise's capability and efficiency to make a smart decision from heterogeneous data sources including structured, semi-structured and unstructured data by developing the cognitive computing technology . The ultimate goal is to provide enterprise and human decision supporting tools for complex problems solving. In this talk, I will brief IBM research strategy on cognitive computing and several technologies we are working on to interact with data naturally and understand data deeply.
 +
 
 +
Advanced Language Modeling
 +
 
 +
•      Presenter:
 +
–      Songfang Huang/黄松芳
 +
–      Research Staff Member
 +
–      IBM Research – China
 +
•      Abstract:
 +
Traditional N-gram models are widely used in speech and language processing applications, e.g., speech recognition and machine translation. However, N-gram models suffer from some limitations, due to the Markov assumption. In this talk, we will briefly review several advanced language modeling techniques to go beyond short-span limitations and incorporate additional information sources. The models we will cover include, but not limited to, exponential models, Bayesian models, and neural network models.
 +
 
 +
Multi-factor Mobile Biometric Authentication
 +
•      Presenter:
 +
–      Min Li/李敏
 +
–      Staff Researcher Member
 +
–      IBM Research - China
 +
•      Abstract:
 +
–      Mobile business transactions are undergoing a surging growth as the raise of mobile Internet and mobile e-Commerce. Security becomes the top concern of mobile customers and enterprises because of potential risks from lost/stolen or unattended mobile devices. This talk will introduce background of mobile money, IBM's mobile biometric authentication solution and some technical progresses.

2014年10月20日 (一) 09:58的最后版本

IBM scientists Yong Qin、Songfang Huang、Ming Li visit CSLT

Time: 2014-05-26
Location: ROOM 1-305, BLDG FIT, Tsinghua University

Cognitive Computing

• Presenter: – Kelvin Qin/秦勇 – Senior Technical Staff Member – IBM Research – China • Abstract: In the era of Big Data, the biggest challenge for enterprises today is to re-discover the trends, re-discover the clients, re-define the relationship with customers to satisfy their ever growing demands. We strive to improve enterprise's capability and efficiency to make a smart decision from heterogeneous data sources including structured, semi-structured and unstructured data by developing the cognitive computing technology . The ultimate goal is to provide enterprise and human decision supporting tools for complex problems solving. In this talk, I will brief IBM research strategy on cognitive computing and several technologies we are working on to interact with data naturally and understand data deeply.

Advanced Language Modeling

• Presenter: – Songfang Huang/黄松芳 – Research Staff Member – IBM Research – China • Abstract: Traditional N-gram models are widely used in speech and language processing applications, e.g., speech recognition and machine translation. However, N-gram models suffer from some limitations, due to the Markov assumption. In this talk, we will briefly review several advanced language modeling techniques to go beyond short-span limitations and incorporate additional information sources. The models we will cover include, but not limited to, exponential models, Bayesian models, and neural network models.

Multi-factor Mobile Biometric Authentication • Presenter: – Min Li/李敏 – Staff Researcher Member – IBM Research - China • Abstract: – Mobile business transactions are undergoing a surging growth as the raise of mobile Internet and mobile e-Commerce. Security becomes the top concern of mobile customers and enterprises because of potential risks from lost/stolen or unattended mobile devices. This talk will introduce background of mobile money, IBM's mobile biometric authentication solution and some technical progresses.