Eventil - Find Tech Events
Official mobile app
FREE - In Google Play
Official mobile app
FREE - In App Store

DSPT#63-The time has come: LSTM pitfalls, will you be the outlier? (Braga)

Oct 15, 2019 · Braga, Portugal

Join us at WeDO Technologies for the next DSPT Braga Meetup to hear about a real case scenario on how to detect outliers using different statistical and ML approaches with Raphael Espanha from WeDO Technologies, and to learn a bit more about LSTM and some of its pitfalls with Fábio Silva from Polytechnic Institute of Porto.

=== SCHEDULE ===
The preliminary agenda for the meetup is the following:
• 18:30-19:00: Welcome and get together
• 19:00-19:30: Talk 1: "Outlier detection: from statistical analysis to more advanced ML/DL approaches"; by Raphael Espanha – WeDO Technologies.
• 19:40-19:45: Group photo
• 19:45-20:15: Networking / Coffee Break
• 20:15-20:45: Talk 2: "Long Short-Term Memory Networks applied in Time Series Machine and Learning Problems"; by Fábio Silva - Polytechnic Institute of Porto (ESTG)
• 20:50: Closing, hanging out and some beers
• 21:00: Dinner is optional but it might be an excellent opportunity for networking (register here: https://doodle.com/poll/6sxa5b62m2rbftmy)

This meetup is sponsored by WeDO Technologies
(https://www.wedotechnologies.com/). Thank you for the support!
Do you want to be a sponsor in future meetups? Please contact us to
See you there!

Talk 1:
Title: Outlier detection: from statistical analysis to more advanced ML/DL approaches

Abstract : A real case scenario will be presented where the goal is to detect ranges of numbers with abnormal behavior. It will be described how the problem was tackled using simple statistical analysis, PCA (Principal Component Analysis) and unsupervised Deep Learning approaches like Autoencoder (AE) and stacked LSTM (Long short-Term Memory) Autoencoder.

Short Bio : With a degree in Biomedical Engineering and a master's degree in Medical Informatics from the University of Minho, Raphael started his internship at Philips in Aachen, Germany. He did his master's thesis at the Cologne University Hospital where he combined medical images and clinical data into a Deep Learning (DL) model, and his work was published in the paper "Combining Image and Non-image Clinical Data: An Infrastructure that Allows Machine Learning Studies in a Hospital Environment". In Portugal, he continued his work at Neadvance on image segmentation projects.
In 2018, he shifted from healthcare to telecom. Currently he is a data scientist at WeDo Technologies where he is developing ML solutions to fight telecom fraud.

Talk 2:
Abstract: The promise of Long Short Term Memory (LSTM) networks is that the temporal dependence and contextual information in the input data can be learnedand generalised to produce reliable predictions. However, literature is evasive in regard to several aspects of the conceived LSTMs and often exhibits misconceptions that maylead to important pitfalls. In this talk, tradional approaches and popular algorithms are compared to LSTM networks in timeseries problems in real world case studies. The pitfalls are analysed by the intrepertation of results and pitfalls of both approaches.

Short Bio: Fábio Silva is an Adjunct Professor at School of Technology and Management in the Polytechnic Institute of Porto, where he lectures classes in the field of Informatics. He is also a researcher in the CIICESI research group of the Polytechnic Institute of Porto and a collaborator of the ALGORITMI Centre research group at University of Minho. During his research activities, he had the opportunity to collaborate in national and European research projects with collaborations with researchers across several countries. His current research interests include Citizen-Centric Smart Cities Services, Computational Sustainability, Internet of Things, Machine Learning and Intelligent Systems.

Event organizers

Are you organizing DSPT#63-The time has come: LSTM pitfalls, will you be the outlier? (Braga)?

Claim the event and start manage its content.

I am the organizer

based on 0 reviews