Date: 23 April 2024
Time: 11:00-12:00 CEST
We unfortunately have to cancel this webinar. It will be rescheduled later. If you had registered for this webinar, we will inform you when we have the new date.
NEASQC is organising a series of interactive webinars to share our findings with the quantum computing community. Each month NEASQC experts present an Open Source library available in the NEASQC GitHub or a technique they investigated, and answer your questions.
In April, we will present some outcomes of the NEASQC use case on Quantum Natural Language Processing.
Abstract: Natural Language Processing (NLP) has been a core focus area of industry over the last decade. Thus, there was a will to seek quantum solutions to NLP tasks. Two major challenges stood on the way to a hybrid NISQ quantum <> classical model for such tasks. First, the sheer amount of data to be handled, realistic NLP datasets being made of tens if not hundreds of thousands of sentences which required adequate scaling of the algorithm. The second being the dimensionality of said data, each sentence represented by high-dimensional vectors, hard to fit on NISQ devices with low qubit count.
In this talk we present a simple light-weight hybrid NISQ-quantum<>classical model to address sentiment analysis and topic classification, two major applications of NLP. Here, the quantum circuit acts as a hidden layer, sandwiched between two multi-layer perceptrons which perform dimensionality reduction as pre-processing and classification as post-processing.
Speakers: Richard Wolf, ICHEC and Mārcis Pinnis, TILDE
Richard Wolf is a Research Fellow at the Irish Center for High-End Computing, co-located with the University of Galway, Ireland. Graduate of the Institut Polytechnique de Paris in France, his master thesis, prepared under the supervision of Bob Coecke, focused on near term applications of QNLP. With a background in linguistics and classical machine learning and algorithms, his core research interests include quantum machine learning and socio-anthropological aspects of the field, as well as quantum error correction.
Mārcis Pinnis is the Chief AI Officer of Tilde, a company developing language technologies. He holds a PhD in Computer Science (Dr.sc.comp.) from the University of Latvia. Mārcis’ primary research focuses on neural machine translation and practical applications of large language models.
Related deliverable and Open Source SW
Events 
News 
