BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Women@CL talklet event - Sheharbano Khattak\, Youmna Farag\, Anita
  Verő
DTSTART:20161124T130000Z
DTEND:20161124T140000Z
UID:TALK68500@talks.cam.ac.uk
CONTACT:Marwa Mahmoud
DESCRIPTION:*Speaker*: Sheharbano Khattak\n\n*Title*: Characterization of 
 Internet Censorship from Multiple Perspectives\n\n*Abstract*: \nCensorship
  of online communications threatens principles of openness and freedom of 
 information on which the Internet was founded. In the interest of transpar
 ency and accountability\, and more broadly to develop scientific rigour in
  the field\, we need methodologies to measure and characterize Internet ce
 nsorship.  Such studies will not only help users make informed choices ab
 out information access\, but also illuminate entities involved in or affec
 ted by censorship\; informing the development of policy and enquiries into
  the ethics and legality of such practices. However\, measurement of Inter
 net censorship is more complex than typical communication network measurem
 ents because of the inherently adversarial and opaque landscape in which i
 t operates. As details about mechanisms and targets of censorship are usua
 lly undisclosed\, it is hard to define exactly what comprises censorship\,
  and how it operates in different contexts. My research aims to help fill 
 this gap by developing three methodologies\, that are then applied to real
 -world datasets to characterize Internet censorship from multiple perspect
 ives\; uncovering entities involved in censorship and targets of censorshi
 p\, and the effects of such practices on stakeholders. In this talk\, I wi
 ll provide an overview of the key contributions\, results\, and impact of 
 my PhD thesis.\n\n------\n\n*Speaker*: Youmna Farag\n\n*Title*: Convolutio
 nal Neural Networks for Automated Essay Assessment\n\n*Abstract*:\nThe tas
 k of assessing students' essays has traditionally relied on human graders 
 and their judgement of writing quality. There has been several attempts to
  build systems that can automate this task to make it more cost- and time-
 efficient. Most of these systems have heavily relied on handcrafted textua
 l features to discriminate between well and poorly written essays. Since e
 xtracting such features is a daunting process\, we propose to employ neura
 l networks to the task of essay scoring. The networks operate on simple wo
 rd vector representations\, hence\, avoiding expensive feature engineering
 . We\, particularly\, apply Convolutional Neural Networks (CNNs).\nCNNs ha
 ve played a fundamental role in the recent breakthroughs in computer visio
 n and object identification. The networks use image pixels and multiple le
 vels of abstraction to effectively identify objects in images. We apply th
 e same idea to predict essay scores by building networks that convolves ov
 er the different textual units: characters\, words and sentences. Our resu
 lts are promising and indicative of the ability of CNNs to evaluate writin
 g quality.\n\n------\n\n*Speaker*: Anita Verő\n\n*Title*: Questions in mu
 lti-modal semantics\n\n*Abstract*:\nIn multi-modal semantics we aim to gro
 und meaning in perceptual input\, usually using images and high performing
  deep visual representations\, learned by convolutional neural networks. H
 owever\, we still have many open questions about the sources and models we
  use\, such as whether: 1) the choice of the network architecture affect p
 erformance\, 2) the difference between search engines and manually annotat
 ed data sources has an impact\, 3) the number of images for each word matt
 er and 4) whether these findings extend to different languages? In my tal
 k I will present our results investigating these issues.\nIn the second pa
 rt I will talk about my next question: whether multi-modality helps with t
 he compositional representation of phrase and sentence meaning.\n
LOCATION:Computer Laboratory\, William Gates Building\, Room FW26
END:VEVENT
END:VCALENDAR