| More Information: | Fully-funded Research Support Officer / PhD Candidate
to conduct research on multimodal semantic grounding and model transparency, at
the interface between Natural Language Processing and Computer Vision.
This project forms part of the EU-funded NL4XAI Innovative Training Network, and
this specific call targets ESR5 in this ITN.
The researcher will work under the supervision of Albert Gatt
(https://staff.um.edu.mt/albert.gatt/) at the University of Malta, and the
co-supervision of Kees van Deemter (
https://www.uu.nl/staff/CJvanDeemter) at the University of Utrecht.
The NL4XAI Training Network focuses on the challenge of making AI systems
intelligible through the use of Natural Language. In this specific project
(ESR5), the focus will be on models -- especially neural models -- which
generate language from perceptual (e.g. image or video) input (for example, to
answer questions, or generate descriptive or narrative captions). Such systems
raise a number of important questions, such as:
(a) to what extent are the choices made by a language model truly grounded in
sensory information?
(b) what is the best way to design training data for such models to ensure that
perceptual and linguistic modalities are complementary, rather than redundant?
(c) how can the choices made by such models be rendered explicit or intelligible
to human users?
These questions will be addressed in a variety of ways, including through
techniques to test the sensitivity of neural Vision-Language models, as well as
behavioural techniques to evaluate them against human input.
Both Albert Gatt and Kees van Deemter have long-standing track records in
research in Natural Language Generation. The successful candidate will be part
of the Institute of Linguistics and Language Technology at the University of
Malta, working in a group of researchers working on several aspects of NLP, as
well as on language from theoretical, experimental and applied perspectives.
Successful candidates will also benefit from the training network which is being
set up under the European NL4XAI Innovative Training Network
(https://nl4xai.eu/).
In the course of their work, the researcher will carry out two three-month
secondments to the University of Utrecht, and one three-month secondment to
Orange in Lanion, France (with Lina Rohas-Barahona and J. Heinecke).
FURTHER INFORMATION:
Applications can be made directly at: https://nl4xai.eu/vacancies/
The full text of the call, with details of employment conditions at the
University of Malta is available here: http://bit.ly/semgrounding
Inquiries should be sent directly to Albert Gatt (albert.gatt@um.edu.mt)
The NL4XAI project has received funding from the European Union’s Horizon 2020
research and innovation programme under the Marie Skłodowska-Curie grant
agreement No 860621. |