Abu Dhabi Machine Learning Season 1 Episode 4
13.04.2021 - Abu Dhabi Machine Learning - ~2 Minutes
When?
- Tuesday, April 13, 2021 from 6:30 PM to 8:30 PM (Abu Dhabi Time)
Where?
- At your home, on zoom. All meetups will be online as long as this COVID-19 crisis is not over.
Programme:
Talk 1: pip install explainerdashboard
Speaker: Oege Dijk, on his explainerdashboard
This package makes it convenient to quickly deploy a dashboard web app that explains the workings of a (scikit-learn compatible) machine learning model. The dashboard provides interactive plots on model performance, feature importances, feature contributions to individual predictions, “what if” analysis, partial dependence plots, SHAP (interaction) values, visualisation of individual decision trees, etc.
GitHub: https://github.com/oegedijk/explainerdashboard
pip install explainerdashboard
Talk 2: AlphaEvolve: A Learning Framework to Discover Novel Alphas in Quantitative Investment
Speaker: Can Cui, PhD student the National University of Singapore
Abstract:
In this talk, Can Cui will present his recent paper on learning to mine alphas. His work is inspired by Google’s AutoML-Zero, and the real-life alphas invested by hedge fund WorldQuant LLC. Authors introduce a new type of alphas combining the strengths of formulaic alphas and machine learning alphas, and then propose a framework to discover those alphas.
paper: AlphaEvolve: A Learning Framework to Discover Novel Alphas in Quantitative Investment
Talk 3: Self-Supervised Learning for Medical Image Analysis
Speaker: Dwarikanath Mahapatra, Senior Research Scientist, Inception Institute of Artificial Intelligence
Abstract:
In this talk I will describe my work on using self-supervised learning (SSL) for medical image analysis. In the absence of labeled data SSL has shown tremendous gains for many deep learning applications such as image classification and segmentation. I will highlight applications of SSL to medical image augmentation, stain normalization and registration of histopathology images. The results highlight the gains achieved in using SSL for medical image analysis tasks and the potential to achieve better performance than state of the art methods.