ONE (1) research grant for candidates with MSc degree with reference number BI|2024/527 is now available under the scope of project ACCELERAT.AI – refª C644865762-00000008

Public notice for research grant

Project ACCELERAT.AI – refª  C644865762-00000008

BI|2024/527

INESC-ID - Instituto de Engenharia de Sistemas e Computadores, Investigação e Desenvolvimento em Lisboa is a R&D institute dedicated to advanced research and development in the fields of Information Technologies, Electronics, Communications, and Energy. INESC-ID has participated in more than 50 research projects funded by the European Union and more than 190 funded by national entities. Until today, our researchers have published more than 700 papers in international journal papers, more than 3000 papers in international conferences, and have registered 15 patents and/or brands.

1 | RESEARCH GRANT TYPE

ONE (1) research grant for candidates with MSc degree with reference number BI|2024/527 is now available under the scope of project ACCELERAT.AI – refª  C644865762-00000008 funded by Recovery and Resilience Plan (RRP)  https://recuperarportugal.gov.pt/ and by european funds Next Generation EU, under the following conditions:

2 | DURATION

FOUR (4)  months, starting in May 2024

-     Renewable, if the candidate is enrolled in a PhD program -  art. 6º, n.4 c)

https://files.dre.pt/2s/2019/12/241000000/0009100105.pdf

subject to suitable performance within the period of the project, not exceeding the maximum period set by FCT for such grants – 4 years (included contract renewals)

-     Renewable, if the candidate is enrolled in a non-degree programme – art. 6º, n. 4 a)

https://files.dre.pt/2s/2019/12/241000000/0009100105.pdf  

subject to suitable performance within the period of the project, not exceeding the maximum period set by FCT for such grants – 1 year (included contract renewals)

3 | LEGISLATION

A fellowship contract will be celebrated according to:

1. Law 40/2004 of 18^th of August (Scientific Research Fellow Status) and its successive amendments, including the amendments introduced by the Decree Law n. 123/2019 of 28 ^th of August 
   https://dre.pt/web/guest/legislacao-consolidada/-/lc/124281176/201912061112/73740605/diploma/indice?lcq=estatuto+do+bolseiro,
    
2. Regulations for Research Grants of the Foundation for Science and Technology in force  https://files.dre.pt/2s/2019/12/241000000/0009100105.pdf

1. INESC-ID Lisboa Grant Regulations

https://www.inesc-id.pt/scholarship-regulations/

1. Recovery and Resilience Plan (RRP)

https://recuperarportugal.gov.pt/

The fellowship contract is awarded on an exclusive dedication basis – art. 5 of Scientific Research Fellow Status and art. 16 of Regulations for Research Grants of the Foundation for Science and Technology.

4 | MONTHLY AMOUNT

The monthly amount of the grant 1 259,64€ is in accordance with the values stipulated in the “Regulations for Research Grants of the Foundation for Science and Technology” in force https://www.fct.pt/wp-content/uploads/2024/02/Tabela-de-Valores-SMM_atualizacao-2024.pdf  and shall be rendered through a monthly bank transfer to an account held by the grantee.

5 | OBJECTIVES/WORKPLAN

A significant number of transformer-based language models specifically tailored for European Portuguese have been recently proposed, such as Albertina, Gervásio or Glória, among others.  These type of models have shown exceptional modelling capabilities of language (understanding and generation) and remarkable performance on a wide range of natural language processing tasks. Concurrently, a similar effort by the research community has resulted in the proposal of several foundation models for speech and audio. These models are trained on extensive amounts of multilingual data to acquire robust representations that capture the intrinsic structure and relationships within speech and audio signals.

The goal of this project is to delve into the existing Portuguese language model landscape, conducting comparative analyses and assessing their potential utility. The final aim is to integrate these models into an automatic speech recognition system. This system will employ a pre-trained speech encoder, one of the studied pre-trained large language models (utilizing either encoder-decoder or decoder-only configurations), and a neural adaptor module.  

The work plan includes the following steps:

1. Research existing foundation models for European Portuguese, both text and speech/audio.

2. Analyse and compare them in common benchmarks.

3. Assess their potential utility as a part of an ASR pipeline that integrates both pre-trained speech encoders and LLMs.

6 | SCIENTIFIC SUPERVISION

The activity will be supervised by Alberto Abad and João Fernando Ferreira, both are researchers at INESC-ID and Associate Professors at Instituto Superior Técnico.

INESC ID will integrate the grantee in the research team of the scientific advisors.

7 | ADMISSION REQUIREMENTS

The candidates should have a MSc in Computer Science and Engineering, Electrical Engineering or Data Science.

By the grant start date, the candidate must be enrolled in

1. a PhD programme – art. 6º, n.1

https://files.dre.pt/2s/2019/12/241000000/0009100105.pdf

or

1. a non-degree programme – art. 6º, n. 2

https://files.dre.pt/2s/2019/12/241000000/0009100105.pdf

Preferential factors:

Preference will be given to candidates who have :

                - Fluent Portuguese speaker;

                - Hands-on experience with popular ML tools, such as TensorFlow, Keras and PyTorch;

                - Knowledge and experience in NLP tasks, in particular, with the most popular LLM architectures.

8 | EVALUATION CRITERIA AND COMMITTEE

The selection will be according to the following criteria:

1. CV (50%)
2. knowledge in the fields of machine learning (20%)
3. knowledge in the specific field of NLP (20%)
4. English language (10%)

An interview can be set up. In this case, the interview accounts for 100% of the criteria.

The jury may also decide not to assign the scholarship, if none of candidates meets the required conditions

9 | COMPLAIN AND APPEAL DEADLINES AND PROCEDURES

The jury has the faculty not to select a candidate who does not prove the requirements mentioned in required education Level and research experience

The admitted and excluded candidates will be notified by email of the final ranking list, including the copy of the Preliminary Report of the jury.

Prior Hearing and Deadline for Final Decision: After being notified, candidates have 10 working days to submit, if applicable, a formal rebuttal.

After that period, the jury notifies the candidates of the Final Report.

Excluded applicants may complain about the jury's final report for 15 working days after notification or appeal the jury's decision to the INESC ID Board of Directors for 30 working days after notification.

According to the Portuguese Law, a disabled candidate has a preference when in equal classification, which prevails over any other legal preference. Candidates must declare their respective degree of disability, the type of disability and the means of communication / expression to be used in the selection process, under the law.

10 | FORMALISATION OF APPLICATIONS

11 | Application Dates