Presentation: Interpretable Machine Learning Products
Share this on:
Abstract
Interpretable models are easier to improve. Regulators and society can better trust them to be safe and nondiscriminatory. They can also offer insights that can be used to change real-world outcomes for the better. But because there is a central tension between accuracy and interpretability interpretability can be hard to ensure.
I'll explore both the product case for interpretability and the academic research that is starting to make the inner workings of black box models such as deep neural networks easier to understand. In particular, I'll look at the application of a new open source tool called LIME to customer churn, image classification and black box NLP models.
Tracks
-
Deep Learning Applications & Practices
Deep learning lessons using tooling such as Tensorflow & PyTorch, across domains like large-scale cloud-native apps and fintech, and tacking concerns around interpretability of ML models.
-
Predictive Data Pipelines & Architectures
Best practices for building real-world data pipelines doing interesting things like predictions, recommender systems, fraud prevention, ranking systems, and more.
-
ML in Action
Applied track demonstrating how to train, score, and handle common machine learning use cases, including heavy concentration in the space of security and fraud
-
Real-world Data Engineering
Showcasing DataEng tech and highlighting the strengths of each in real-world applications.
-
Handling Sequential Data Like an Expert / ML Applied to Operations
Discussing the complexities of time (half track) and Machine Learning in the data center (half track). Exploring topics from hyper loglog to predictive auto-scaling in each of two half-day tracks.
Half-day tracks -
AI Meets the Physical World
The track where AI touches the physical world. AI use cases around drones, self-driving cars, ROS, NVidia Jetson, & Amazon Deep Lens.