PyData Meetup @ Prodrive Technologies

It is time for another edition of the PyData Eindhoven meetup. Join us on the 26th of April at Prodrive Technologies for an evening of technical talks for technical people. Please note that this event will take place at Prodrive Technologies and there will be a tour of their facilities.

We were humbled by the large interest at the PyData Conference in December 2022. To keep the momentum going we are organising this event around the theme: Python and Mechatronics.

Agenda:
17:30 - Doors open
17:45 - welcome (from PyData, announcements)
17:50 - welcome from Prodrive Technologies
18:00 – Tour of Prodrive facilities
18:45 - Free Dinner
19:30 – Getting Started with Anomalib – Carlos from ING (https://www.linkedin.com/in/carlosdelarosa1/)
20:00 – Automatic Noise Filtering with Dynamic Sparse Training in Deep Reinforcement Learning – Bram from TU/e (https://www.linkedin.com/in/bramgrooten/)
20:30 – Drinks and networking
----
Anomalies detection with anomalib Anomalies detection is the data-driven task of identifying these rare occurrences and filtering or modulating them from the analysis pipeline and trying to identify correlations with these events that can be used on financial fraud, equipment fault, or irregularities on time series analysis. Anomalib is an open-source library that allows anomaly detection ready-to-use deep learning with the largest public collection of datasets focus on image -base anomalies detection using technologies like pytorch lighting and openVINO for accelerated inference on intel hardware. We will be getting started checking the architecture of the process and setup of the environment to start using the model for analysis.
----
Title of talk:
Automatic Noise Filtering with Dynamic Sparse Training in Deep Reinforcement Learning

Short description:
Tomorrow's robots will need to distinguish useful information from noise when performing different tasks. A household robot for instance may continuously receive lots of information about the home, but needs to focus on just a small subset to successfully execute its current chore. Filtering distracting inputs that contain irrelevant data has received little attention in the reinforcement learning literature. To start resolving this, we formulate a problem setting in reinforcement learning called the extremely noisy environment (ENE), where up to 99% of the input features are pure noise. Agents need to detect which features provide task-relevant information about the state of the environment. Consequently, we propose a new method termed Automatic Noise Filtering (ANF), which uses the principles of dynamic sparse training in combination with various deep reinforcement learning algorithms. The sparse input layer learns to focus its connectivity on task-relevant features, such that ANF-SAC and ANF-TD3 outperform standard SAC and TD3 by a large margin, while using up to 95% fewer weights. Furthermore, we devise a transfer learning setting for ENEs, by permuting all features of the environment after 1M timesteps to simulate the fact that other information sources can become relevant as the world evolves. Again, ANF surpasses the baselines in final performance and sample complexity.

Links to appear:
Personal webpage: https://www.bramgrooten.nl/
Paper: https://arxiv.org/abs/2302.06548
PyTorch code: https://github.com/bramgrooten/automatic-noise-filtering
If there is enough space you may also add:
LinkedIn: https://www.linkedin.com/in/bramgrooten/
Twitter: https://twitter.com/BramGrooten
Google Scholar: https://scholar.google.com/citations?user=zkYA_KEAAAAJ
----
Attendee profile:

• People who want to learn about Python, Julia, R in the context of Data
• Interested in seeing how others use Python for solving real problems
• Interested in being part of the Python Community in Eindhoven
• No Recruiters
• No Sales
• Only people who speak Data