Title:

Searching  for Solutions: Uses of Meta-heuristic Search in Security

Lecturer:

Professor  John A Clark

Time:

14:30, September  12, 2016

Venue:

1012, New  Science & Technology Building, North Campus

Lecturer  Profile

Professor  Clark holds a Master of Arts in Mathematics and a Master of Science in Applied  Statistics. He joined York in 1992 as Lecturer in Safety Critical Systems and  has been research active since around 1997. His personal and supervised  research work concentrates on aspects of security and software engineering,  with a particular interest in applying metaheuristic search and other  Artificial Intelligence techniques to these areas. He was awarded a PhD in Computer  Science in 2002 (on applications of metaheuristic search to cryptology) and  promoted to a Personal Chair in 2005. His research work has been cited over  5580 times. He is part of the EPSRC's Programme Grant on Dynamic Adaptive  Automated Software Engineering (DAASE). His collaborative work has been awarded  11 best paper (or similar) prizes. In 2013 he received a personal Royal Society  Wolfson Research Merit Award to pursue his work in optimisation based system  design and analysis.

He has maintained a very strong interest in  teaching throughout his academic career. He has taught on around 30 modules  since joining York. From 2005-2007 he was Chair of Examinations within the  Department of Computer Science at York. From 2009-2015 he was the Deputy Head of  Department, Responsible for Research. From October 2016 he will be the Head of  Department.

Lecture  Abstract

Meta-heuristic  search has sought to use general purpose search and optimisation strategies to  navigate complex design and analysis spaces. The problem in question is  generally expressed in the following form: given a search space X, find x in X,  such that some fitness function f(x) is maximised (or minimised). In this talk  Professor Clark will show how optimisation approaches such as simulated annealing  and evolutionary computation approaches (genetic algorithms, genetic  programming and others) can be used to design secure artefacts such as  cryptographic components and security protocols. The presentation will also  show how some types of security schemes can be broken by using such  techniques. Finally the use of such approaches to discover ideas for analysis  strategies will be discussed.

Title:

Verifiable  Databases with Efficient Updates in Cloud Computing

Lecturer:

Professor  CHEN Xiaofeng

Time:

15:05,  September 12, 2016

Venue:

1012, New  Science & Technology Building, North Campus

Lecturer  Profile

Xiaofeng  Chen received his B.S. and M.S. on Mathematics from Northwest University, China  in 1998 and 2000, respectively. He got his Ph.D degree in Cryptography from  Xidian University in 2003. Currently, he works at Xidian University as a  professor. His research interests include applied cryptography and cloud  computing security. He has published over 100 research papers in refereed  international conferences and journals. His work has been cited more than 4000  times at Google Scholar. He is in the Editorial Board of IEEE transactions on  Depenable and Secure Computing (TDSC)，Security and Communication Networks (SCN), Computing and Informatics  (CAI), and International Journal of Embedded Systems (IJES) etc. He has served  as the program/general chair or program committee member in over 30  international conferences.

Lecture  Abstract

The notion  of verifiable database (VDB) enables a resource-constrained client to securely  outsource a very large database to an untrusted server so that it could later  retrieve a database record and update a record by assigning a new value. Also,  any attempt by the server to tamper with the data will be detected by the client.  In this talk, we introduce some recent progress on VDB schemes.

Title:

Syntax  based word embeddings for improved natural language processing (and overview of  AI Group research)

Lecturer:

Dr. Suresh  Manandhar

Time:

15:55,  September 12, 2016

Venue:

1012, New Science & Technology Building,  North Campus

Lecturer  Profile

Dr Manandhar received his MSc in Artificial  Intelligence from University of Essex, UK in 1987 and his PhD in Artificial  Intelligence from University of Edinburgh, UK in 1994. He has a wide range of  interests in Natural Language Processing focusing on unsupervised learning of  semantics, morphology, applications of deep learning and on question answering  systems. Dr Manandhar has published over 100 peer reviewed papers and successfully  supervised 15 PhD students. He currently serves on the editorial boards of the  Journal of Natural Language Engineering and the Journal of Applied  Intelligence.

Lecture  Abstract

I  will provide a brief overview of the research being undertaken by the AI Group  at York. I will follow this by the current work being undertaken in natural  language processing (NLP) focussing on word embeddings. Word embeddings are low  dimensional vector space meaning representations for words. There has been a  huge interest in the use of word embeddings for many applications in NLP.   In this talk, a variation to standard word embeddings will be presented that  enable capturing syntactic information and allowing the word embeddings to be  modified dynamically based on the context. In particular, dependency syntax is  employed. Three common sentence classification tasks: question type  classification on the TREC, dataset, binary sentiment classification on  Stanford’s Sentiment Treebank and semantic relation classification on the SemEval  201 dataset. Evaluation is done using three different classification methods: a  Support Vector Machine, a Convolutional Neural Network and a Long Short Term  Memory Network. This works shows that dependency based embeddings outperform  standard window based embeddings in most of the settings.

Our embeddings and code are available at http://www.cs.york.ac.uk/nlp/extvec

Title:

A New  Ranking Based EA for MaOPs

Lecturer:

Professor WANG  Yuping

Time:

16:30, September  12, 2016

Venue:

1012, New  Science & Technology Building, North Campus

Lecturer  Profile

Prof.  Yuping Wang received his Ph.D. degree in computation mathematics from Xi’an  Jiaotong University, Xi’an, China, in 1993. From 1997 to 2010, he was a  visiting scholar with the Chinese University of Hong Kong, the Hong Kong  Baptist University and the City University of Hong Kong, respectively. He  served as a Chair or Co-Chair of several international conferences, Guest  Editor of several International Journals and a member of Editorial Board of  Journal: ICAE. He is currently a full professor with Xidian University, Xi’an,  China and a senior member, IEEE. He has authored and co-authored over 100  research papers in international journals and conferences. His current research  interests include evolutionary computation, data mining, optimization  algorithms and modeling for engineering.

Lecture  Abstract

Many  objective optimization problems (MaOPs) are the most challenging problems among  multi-objective optimization problems. In this talk, a new ranking method is  first proposed, then, a new evolutionary algorithm is proposed based on it.  Moreover, the computer simulations are conducted and the results indicate the  proposed algorithm is efficient and effective.

Title:

Computer  assisted safety argument review, a dialectics approach

Lecturer:

Dr. Tommy  Yuan

Time:

17:05,  September 12, 2016

Venue:

1012, New  Science & Technology Building, North Campus

Lecturer  Profile

Professor  Tommy Yuan received his BSc on Railway Transport Engineering from Southwest  Jiaotong University, China in 1993, and his MSc on Software Engineering, and PhD  on Computer Science from Leeds Metropolitan University, UK. He worked for 4  years in assistant professorship in an Icelandic University and 5 years in  Chinese industry as a railway transport engineer and railway station manager.  Since 2009, he joined University of York and has been working as the lecturer  of Artificial Intelligence Research Group in the Department of Computer Science  and Chinese University Collaboration Director. His research concerns  argumentation technology and its applications in technology enhance learning  (TEL), assistive technologies for autism children and system safety case  development. I am also interested in applying artificial intelligent  technologies e.g. machine learning and pattern recognition, to industrial  applications such as transport systems. He has published over 40 conference and  journal papers in the area of artificial intelligence and argumentation.

Lecture  Abstract

There has  been increasing use of argument-based approaches in the development of safety  critical systems. Within this approach, a safety case plays a key role in the  system development life cycle. The key components in a safety case are safety  arguments, which are designated to demonstrate that the system is acceptably  safe. Too often, safety arguments are constructed without proper reasoning.  Inappropriate reasoning in safety arguments could undermine a system's safety  claims, which in turn contributes to safety-related failures of the system. In  this talk, Dr. Yuan will show how the problem can be tackled via an informal  logic based approach. A number safety argument schemes and a safety argument  review model have been proposed and evaluated. It is anticipated that this work  will contribute toward the development of computer system safety arguments, and  help to move forward the interplay between research in informal logic and  research in computer system safety engineering.