School Seminar Series: Statistically Consistent Estimation and Efficient Inference for Natural Language Parsing

Statistically Consistent Estimation and Efficient Inference for
Natural Language Parsing
By Shay Cohen, University of Edinburgh.

Abstract:
In the past few years, there has been an increased interest in the machinel earning community in spectral algorithms for estimating models with latent variables. Examples include algorithms for estimating mixture of Gaussians or for estimating the parameters of a hidden Markov model.

The EM algorithm has been the mainstay for estimation with latent variables, but because it is guaranteed to converge to a local maximum of the likelihood, it is not a consistent estimator. Spectral algorithms, on the other hand, are often shown to be consistent. They are often more computationally efficient than EM.

In this talk, I am interested in presenting two types for spectral algorithms for latent-variable PCFGs, a model widely used in the NLP community for parsing. One algorithm is for consistent estimation of L-PCFGs, and the other is for efficient inference with L-PCFGs (or PCFGs). Both algorithms are based on linear-algebraic formulation of L-PCFGs and PCFGs.

BIO:
Shay Cohen is a Chancellor’s fellow (assistant professor) at the University of Edinburgh (School of Informatics). Before that, he was a postdoctoral research scientist in the Department of Computer Science at Columbia University, and held an NSF/CRA Computing Innovation Fellowship. He received his B.Sc. and M.Sc. from Tel Aviv University in 2000 and 2004, and his Ph.D. from Carnegie Mellon University in 2011. His research interests span a range of topics in natural language processing and machine learning, with a focus on structured prediction. He is especially interested in developing efficient and scalable parsing algorithms as well as learning algorithms for probabilistic grammars.

Event details

  • When: 21st January 2015 14:00 - 15:00
  • Where: Cole 1.33a
  • Series: School Seminar Series
  • Format: Talk

School Seminar: Cloud Platform in Financial Services – Allan Beck, J.P. Morgan

Title: Cloud Platform in Financial Services

Presenter: Allan Beck, Cloud Platform and Strategy Lead from JPMorgan Chase

Abstract: Cloud Computing is revolutionising the delivery of compute services and driving the next generation of web-scale application design. This presents enormous opportunities but also challenges, particularly in heavily regulated sectors such as Financial Services.
Allan Beck, Cloud Platform and Strategy Lead from JPMorgan Chase, will discuss the current approach and challenges to Cloud in Financial Services. This will include an overview of available Cloud services and capabilities, the specific challenges to Cloud in Financial Services (private and public Cloud) and an overview of the next-generation Cloud platform and developer experience at JPMorgan Chase.

Event details

  • When: 18th November 2014 14:00 - 15:00
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Seminar, Talk

Thursday Seminar from Japan – Prof. Yoshifumi Kitamura – Interactive Content Design and 3D Interactions

Date: 2014-11-13
Time: 14:00 to 15:00
Location: Maths Lecture Theatre A, North Haugh, University of St Andrews.

Title: Interactive Content Design and 3D Interactions

Abstract: Good media content has the power to enrich our lives. We focus on non-traditional content other than movies, music and games, conducting comprehensive research on a variety of interactive content which creates new value through interactions with humans. In this talk I will introduce a series of my group’s recent research projects involving technologies for interactive content design and 3D interactions.
Web: http://www.icd.riec.tohoku.ac.jp/index-e.html

Bio: Yoshifumi Kitamura is a Professor at Research Institute of Electrical Communication, Tohoku University. He received the B.Sc., M.Sc. and PhD. degrees in Engineering from Osaka University in 1985, 1987 and 1996, respectively. Prior to Tohoku University, he was an Associate Professor at Graduate School of Engineering and  Graduate School of Information Science and Technology, Osaka University (1997-2010), and before that he was a researcher at ATR Communication Systems Research Laboratories (1992-1996) and Canon Inc. (1987-1992).

Event details

  • When: 13th November 2014 14:00 - 15:00
  • Where: Maths Theatre A
  • Format: Seminar, Symposium, Talk

Distinguished lecture 2014

The first of this academic year’s distinguished lectures will be given by Prof Luca Cardelli of Microsoft Research and the University of Oxford, 0930–1600 on Tuesday 25 November in Lower College Hall.

Continue reading

Event details

  • When: 25th November 2014 09:15 - 16:00
  • Where: Lower College Hall
  • Series: Distinguished Lectures Series
  • Format: Distinguished lecture, Lecture

Supporting the Design of Shape-Changing Interfaces by Jason Alexander, Lancaster University

Abstract:

Shape-changing interfaces physically mutate their visual display surface to better represent on-screen content, provide an additional information channel, and facilitate tangible interaction with digital content. The HCI community has recently shown increasing interest in this area, with their physical dynamicity fundamentally changing how we think about displays. This talk will describe our current work supporting the design and prototyping of shape-changing displays: understanding shape-changing application areas through public engagement brainstorming, characterising fundamental touch input actions, creating tools to support design, and demonstrating example implementations. It will end with a look at future challenges and directions for research.

Bio:

Jason is a lecturer in the School of Computing and Communications at Lancaster University. His primary research area is Human-Computer Interaction, with a particular interest in bridging the physical-digital divide using novel physical interaction devices and techniques. He was previously a post-doctoral researcher in the Bristol Interaction and Graphics (BIG) group at the University of Bristol. Before that he was a Ph.D. student in the HCI and Multimedia Lab at the University of Canterbury, New Zealand. More information can be found at http://www.scc.lancs.ac.uk/~jason/

This seminar is part of our ongoing series from researchers in HCI. See here for our current schedule.

Event details

  • When: 11th November 2014 14:00 - 15:00
  • Where: Cole 1.33a
  • Format: Seminar

School Seminar: Complex Networks and Complex Processes

Simon Dobson, School of Computer Science, University of St Andrews

Abstract:

Complex networks provide a way of modelling systems with lots of
dependent elements, such as traffic networks or social networks. By
running processes over these networks we can explore how the topology of
the network affects the way the process evolves, and potentially
identify factors that accelerate or impede it. This opens-up
possibilities both for study (science) and control (engineering).

This talk will briefly introduce the mechanics of complex networks and
the processes that run on them, review some recent results we have
obtained, and look to future research programme where we will combine
simulation with sensing to give us new ways of looking at the world.

Event details

  • When: 4th November 2014 14:00 - 15:00
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Talk

Emotion Sense: From Design to Deployment by Neal Lathia, Cambridge University.

Abstract:
In the UK, more than 70% of mobile users now own a smartphone. These increasingly powerful, sensor-rich, and personal devices present an immense opportunity to monitor health-related behaviours and deliver digital behaviour-change interventions at unprecedented scale.

However, designing and building systems to measure and intervene on health behaviours presents a number of challenges. These range from balancing between energy efficiency and data granularity, translating between behavioural theory and design, making long psychological assessments usable for end users, and making sense of the sensor and survey data these apps collect in a multidisciplinary setting.

Approximately 18 months ago, we launched Emotion Sense, a mood-tracking app for Android where we tried to address some of these challenges. To date, the app has been downloaded over 35,000 times and has an active user base of about 2,000 people: in this talk, I will describe how we designed, trialled, and launched Emotion Sense, and the insights we are obtaining about diurnal patterns of activity and happiness that we are finding by mining the 100 million+ accelerometer samples the app has collected to date. I’ll close with future directions of this technology — including a novel smoking cessation intervention (Q Sense), and a generic platform (Easy M) that we have developed to allow researchers to conduct their own studies.

http://emotionsense.org/
http://www.qsense.phpc.cam.ac.uk/
http://www.cl.cam.ac.uk/~nkl25/easym/

Bio:
Neal is a Senior Research Associate in Cambridge University’s Computer Laboratory. His research to date falls somewhere in the intersection of data mining, mobile systems, ubiquitous/pervasive systems, and personalisation/ recommender systems, applied to a variety of contexts where we measure human behaviour by their digital footprints. He has a PhD in Computer Science from University College London. More info/contact http://www.cl.cam.ac.uk/~nkl25/

This seminar is part of our ongoing series from researchers in HCI. See here for our current schedule.

Event details

  • When: 28th October 2014 14:00 - 15:00
  • Where: Maths Theatre D
  • Format: Seminar

The Design and Implementation of Feldspar

By: Josef Svenningsson, Chalmers University of Technology, Sweden

Feldspar is a domain specific language with the goal of raising the
level of abstraction for performance sensitive, low-level code.
Feldspar is a functional language embedded in Haskell, which offers a
high-level style of programming. The key to generating generating
efficient code from such descriptions is to use a high-level
optimisation technique called vector fusion. Feldspar achieves
vector fusion for free by employing a particular way of embedding the
language in Haskell by combining deep and shallow embeddings.

Bio: Josef Svenningsson is an Assistant Professor in the Functional
Programming group at Chalmers University of Technology. He has a broad
range of interest and has published papers on wide variety of topics,
including: program analysis, constraint solving, security, programming
language design, testing and high-performance computing.

Event details

  • When: 21st October 2014 14:00 - 20th October 2014 15:00
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Seminar