DSPT#72 - Neural Networks & Evolutionary Algorithms (Leiria)

Do you wanna hear about Neural Networks applied to Natural Language Processing? Or Evolutionary Algorithms?

Happy to announce our next DSPT Meetup on partnership with the TICE.Leiria Community !

=== SCHEDULE ===
• 18:30-19:00: Welcome and get together
• 19:00-19:30: Talk 1: Conversation Topic Classification at Talkdesk
• 19:40-19:45: Group photo
• 19:45-20:15: Networking / Coffee Break
• 20:15-20:45: Talk 2: How I Learned to Stop Worrying and Love Artificial Intelligence
• 20:50: Closing

Talk 1: Conversation Topic Classification at Talkdesk

Abstract: Talkdesk offers a Contact Center in the cloud that handles thousands of calls every day, and for each call we want to know what are the different intents expressed during the conversation (e.g. customer complaining, request for information, scheduling a meeting, etc). In this talk we will show how we are tackling this issue, starting by explaining how we do sentence classification to identify possible changes in the conversation topic, using intent detection techniques that take advantage of Neural Network approaches based on LSTMs. With sentence classification we can then group together sentences related to a specific topic (aka moments). We are also going to show how it is possible to group similar moments using unsupervised methods, such as clustering, and, on top of that, use supervised classification to identify moments that were manually annotated on previous conversations. To achieve this we are using state-of-the-art sentence2vec approaches like BERT, InferSent, Universal Sentence Encoder and others.

Bio: My name is Pedro Verruma and I'm an Engineering Manager at Talkdesk. I've worked in Artificial Intelligence for the past 10 years, most of them using Natural Language Processing techniques to create innovative products for companies like Wizdee, Critical Software and now Talkdesk. Among others, I've specialized in solutions for Knowledge Representation, Semantic Search, Recommendation and Text Classification. I'm also a container deployment and microservices enthusiast with focus on high-availability and high-scalability.

Talk 2: How I Learned to Stop Worrying and Love Artificial Intelligence

Abstract: Nature is a great designer and problem solver. Looking around us it is possible to see endless forms most beautiful that resulted from years and years of constant evolution. Amazed by nature’s ingenious inventive capacity, computer scientists have always tried to understand the way it works, by developing computational methods that can tackle complex problems. Remarkable examples of these proposals are the Artificial Neural Networks and Evolutionary Algorithms. The former are computational models inspired by the brain, whilst the latter are based on the natural selection theory posit by Darwin and inheritance mechanisms proposed by Mendel.
In this talk we will discuss how evolutionary algorithm can help us design an artificial brain that is able to distinguish images, and how they can help us design models that can predict the glucose levels in the blood.

Bio: Nuno Lourenço is an Assistant Professor at the Department of Informatics Engineering of the University of Coimbra, where he obtained is PhD in Information Science and Technology in 2016. He is the coordinator of the Evolutionary and Complex Systems (ECOS) group, and is a member of the Centre for Informatics and Systems of University of Coimbra (CISUC) since 2009. Limited by the text, rest of the extended bio here : https://www.cisuc.uc.pt/people/show/2906