Logic List Mailing Archive

Postdoctoral position in artificial intelligence, Cardiff (Wales), Deadline: 4 Feb 2017

Applications are invited for a Research Associate post in Cardiff 
University?s School of Computer Science & Informatics.

This is a full-time, fixed-term post for 30 months, starting on 1 March 
2017 or as soon as possible thereafter. The successful candidate will be 
dedicated to finding creative solutions and have a genuine curiosity and 
enthusiasm to undertake world-class research in the field of Artificial 
Intelligence. Specifically, the aim of this post will be to develop novel 
methods for learning interpretable/symbolic models from diverse sources of 
information, including knowledge graphs, vector space models and natural 
language text. These models will then be used as background theories in 
applications such as recognising textual entailment, automated knowledge 
base completion, or zero-shot learning. You will work closely with Steven 
Schockaert. You will possess or be near the completion of a PhD in 
Computer Science or a related area, or have relevant industrial 
experience.

This research will be part of the ERC project FLEXILOG

Closing date for applications: 4 February 2017


Job specification:
*******************

Essential criteria:
1.   Postgraduate degree at PhD level, or near to completion of a PhD in a related subject area or relevant industrial experience
Knowledge, Skills and Experience.
2.   An established expertise and proven portfolio of research and/or relevant industrial experience within at least two of the following research fields: Machine Learning, Knowledge Representation, Natural Language Processing.
3.   A strong background in statistics and linear algebra.
4.   Excellent programming skills.
5.   Knowledge of current status of research in specialist field.
6.   Proven ability to publish in relevant journals (e.g. Artificial Intelligence, Journal of Artificial Intelligence Research, Journal of Machine Learning Research, Machine Learning) or top-tier conferences (e.g. IJCAI, AAAI, ECAI, NIPS, ICML, KDD, ACL, EMNLP).
7.   Ability to understand and apply for competitive research funding.
8.   Proven ability in effective and persuasive communication.
9.   Ability to supervise the work of others to focus team efforts and motivate individuals.
10.  Proven ability to demonstrate creativity, innovation and team-working within work.

Desirable criteria:
1.   Experience with learning logical theories from data (e.g. rule learning, taxonomy learning, path ranking algorithms)
2.   Experience with vector space models
3.   Experience with processing large amounts of data using a high-performance computing environment


Background about the project:
**********************************

Vector space embeddings have become a popular representation framework in 
many areas of natural language processing and knowledge representation. In 
the context of knowledge base completion, for example, their ability to 
capture important statistical dependencies in relational data has proven 
remarkably powerful. These vector space models, however, are typically not 
interpretable, which can be problematic for at least two reasons. First, 
in applications it is often important that we can provide an intuitive 
justification to the end user as to why a given statement is believed, and 
such justifications are moreover invaluable for debugging or assessing the 
performance of a system. Second, the black box nature of these 
representations makes it difficult to integrate them with other sources of 
information, such as statements derived from natural language, or from 
structured domain theories. Symbolic representations, on the other hand, 
are easy to interpret, but classical inference is not sufficiently robust 
(e.g. in case of inconsistency) and too inflexible (e.g. in case of 
missing knowledge) for most applications.

The overall aim of the FLEXILOG project is to develop novel forms of 
reasoning that combine the transparency of logical methods with the 
flexibility and robustness of vector space representations. For example, 
symbolic inference can be augmented with inductive reasoning patterns 
(based on cognitive models of human commonsense reasoning), by relying on 
fine-grained semantic relationships that are derived from vector space 
representations. Conversely, logical formulas can be interpreted as 
spatial constraints on vector space representations. This duality between 
logical theories and vector space representations opens up various new 
possibilities for learning interpretable domain theories from data, which 
will enable new ways of tackling applications such as recognising textual 
entailment, automated knowledge base completion, or zero-shot learning.


More information:
********************

For more details about the project and instructions on how to apply, 
please go to www.cardiff.ac.uk/jobs and search for job 5545BR. Please note 
the requirement to evidence all essential criteria in the supporting 
statement.
--
[LOGIC] mailing list
http://www.dvmlg.de/mailingliste.html
Archive: http://www.illc.uva.nl/LogicList/

provided by a collaboration of the DVMLG, the Maths Departments in Bonn and Hamburg, and the ILLC at the Universiteit van Amsterdam