This week in AI & Machine Learning: Adobe AI features, AI Alzheimer’s predictions, Adversarial images and attacks, Learning to Drive with Neural Nets, Feature Stores for MLOps, and more!
Artificial Intelligence News:
This week, Adobe held its 2020 creative conference, Adobe MAX. Among the over 350 sessions and hundreds of speakers, Adobe showed off some really neat new AI-powered features that are available now in the Creative Cloud suite. It’s exciting to see so many cool features utilize neural filters in a single release. Check out the video below to see a preview!
Artificial Intelligence researchers at IBM and Pfizer have been working on a model to predict the onset of Alzheimer’s disease. The AI model uses samples from a verbal test and is reported to have a higher accuracy than current prediction methods.
As machine learning and AI make their way into more fields, addressing bias in the systems has never been more important. In this article, learn about how much the FDA is stressing the need for diversity in datasets–especially when it comes to the healthcare industry. No matter the field, I think data and model bias should be something everyone pays attention to.
Developer Tools & Education:
Last week, we talked about the announcement of pytorch lighting, but this week you can read much more about the launch, features and API.
Learn about creating adversarial images to trick neural networks with tensorflow. Later in this series you will learn how to defend against them.
Deeplearning.ai launches a third course on generative adversarial networks (GANs). Learn more about how to apply GANs in the real world and implement more advanced forms like Pix2Pix.
Interesting Podcasts & Interviews:
- How AI will revolutionize manufacturing | MIT Technology Review
- What’s Next for Fast.ai? with Jeremy Howard | TWiML
- George Hotz: Hacking the Simulation & Learning to Drive with Neural Nets | Lex Fridman Podcast
- Feature Stores for MLOps with Mike del Balso | TWiML
- Chris Lattner: The Future of Computing and Programming Languages | Lex Fridman Podcast
- Why AI Innovation and Social Impact Go Hand in Hand with Milind Tambe | TWiML
Notable Research Papers:
Some of the interesting machine learning papers published this week.
- Generative causal explanations of black-box classifiers
- On the Theory of Transfer Learning: The Importance of Task Diversity
- Not all parameters are born equal: Attention is mostly what you need
- PseudoSeg: Designing Pseudo Labels for Semantic Segmentation
- Why Are Convolutional Nets More Sample-Efficient than Fully-Connected Nets?
- SDF-SRN: Learning Signed Distance 3D Object Reconstruction from Static Images
- Accelerating Reinforcement Learning with Learned Skill Priors
- Few-Shot Adaptation of Generative Adversarial Networks
- Batch Exploration with Examples for Scalable Robotic Reinforcement Learning