picture: André Martins has been awarded a European Analysis Council Consolidator Grant.
Credit score: André Martins
André Martins, a Portuguese researcher from Instituto de Telecomunicações (IT) and CMU Portugal alum, has received a European Analysis Council Consolidator grant (ERC CoG) of 2 M€ to examine synthetic neural networks utilized to pure language processing (NLP).
The ERC Consolidator funding will be utilized to the DECOLLAGE mission (DEep COgnition Studying for LAnguage GEneration), targeted on discovering options to a number of the elementary issues of NLP, utilizing an revolutionary interdisciplinary methodology that brings collectively instruments from synthetic intelligence, sparse modeling, neuroscience, and cognitive sciences.
The prevailing deep studying fashions for pure language processing, whereas generally spectacular, are sometimes unreliable and even deceptive: they don’t generalize properly to new domains, they don’t exploit contextual data, they’re poorly calibrated, and their reminiscence shouldn’t be traceable. These limitations stem from their monolithic architectures, appropriate for notion however unsuitable for duties that require higher-level cognition.
“This project is a step further in overcoming the limitations of current NLP technologies, making it possible for humans and machines to communicate effectively in natural language and to work collaboratively to solve increasingly harder problems”, André Martins
With the funding accessible underneath this new Grant, the analysis will give attention to three principal steps. First, utilizing uncertainty and high quality estimates as a tenet for managed technology, combining this management mechanism with effectively encoding contextual data and integrating a number of modalities. Second, creating sparse and structured fashions of long-term reminiscence, with consideration to descriptive representations, and third, devising new mathematical fashions for sparse communication (bridging the hole between discrete and steady representations), supporting end-to-end differentiability and enabling a shared workspace the place a number of modules and brokers can talk. These improvements will probably be utilized to extremely difficult technology duties, together with machine translation, dialogue, and open-ended technology.
This ERC award follows an early beginning grant received by the researcher in 2017, value 1.4 million euros, utilized to the DeepSPIN mission that allowed him to develop his analysis in deep studying and structured prediction to unravel difficult duties in pure language processing, together with machine translation, high quality estimation, and syntactic parsing.
André Martins was the primary CMU Portugal Twin Diploma Ph.D. Alumni in Language Applied sciences, having concluded his doctoral diploma in 2012 granted by Técnico and Carnegie Mellon College (CMU). Since then, he has been extremely concerned with the Program initiatives, supervising two Ph.D. candidates and main two CMU Portugal Giant Scale tasks (MAIA and GoLocal) in strict collaboration with groups at CMU. He’s at present Vice President of AI Analysis at Unbabel and an Affiliate Professor at Instituto Superior Técnico (Técnico). He co-founded and co-organizes the Lisbon Machine Studying College (LxMLS), already in its twelfth Version, which counts with the assist of the CMU Portugal Program, and is a Fellow of the ELLIS society, the pan-European AI community. With over 100 papers printed, +6,000 citations, and an h-index of 37 in top-tier conferences and journals, André stands among the many prime 2% of the most-cited scientists in his scientific subject by Elsevier.
Disclaimer: AAAS and EurekAlert! should not accountable for the accuracy of stories releases posted to EurekAlert! by contributing establishments or for using any data via the EurekAlert system.