This week in AI & Machine Learning: improving logistics with ML, new drug research methods, spaCy 3.0, Long-Term Network for Long-Term Motion Prediction, and more!
Artificial Intelligence News:
The impacts of COVID-19 are felt almost everywhere in 2020, one field that has seen a massive change is logistics. As more people have switched to ordering online in order to maintain limited exposure to the virus companies have had to adjust the way they think about managing orders and deliveries. This interview with Kapil Bharati, CTO and Co-founder at Delhivery covers some of the ways they embraced machine learning to help optimize their 1 million daily shipments, and all the steps in between.
One area in the healthcare industry that doesn’t seem to have benefited from many new solutions in the marketplace is hearing loss. Learn a little bit more about this field, and how some startups can use machine learning to perform advanced signal processing.
Do you think the hearing aid is something that could be improved with AI or are the current solutions already good enough?
Researchers recently made a breakthrough using machine learning to help improve the energy output from nuclear fusion reactors by simulating the plasma interaction with materials.
Researchers at MIT have started using machine learning to improve drug compound screenings that could help biologists and pharmaceutical companies make better predictions on molecule bindings for development.
Read about some of the ways Sixgill’s platform can be used in drug discovery R&D by the pharmaceutical industry.
Developer Tools & Education:
Pytorch just released version 1.0. Along with the latest release, the official pytorch site got a new look.
spaCy version 3.0 is going to be a big release with plenty of new features. You can get started using 3.0 with the new spaCy nightly builds
I was wondering when this would arrive! Udacity has partnered with Microsoft to launch a comprehensive course on Azure machine learning.
Upcoming Online AI & Data Events:
Join this webinar hosted by AICamp to learn about DeTextt, a state-of-the-art open source NLP framework with BERT/CNN/LSTM encoders for text data processing and understanding.
Introduction to Computer Vision: How to build an object detection model with your own dataset | 10/20 – 5:30pm PDT
Join me for this hands on workshop to learn the basics of building a deep learning object detector and how to label your own dataset in a practical way.
Join this webinar hosted by pytorch lightning for an AMA with their CEO and CO-Founder, William Falcon.
Join this workshop to learn the theoretical concepts for deep learning, and a hands-on code lab.
Interesting Podcasts & Interviews:
Listen to the latest conversation with Scott Aaronson and Lex Fridman to hear their discussion around simulation, the theories of everything, consciousness, turing tests, GPT-3, and much more.
Learn about strengthening machine learning and artificial intelligence in African, the distinction between decolonizing AI and ethical AI, and explore the origin of deep learning Indaba.
Learn about origins of the MLT community, Suzana’s work at Causaly and transition from linguist and domain expert to working with causal modeling.
Notable Research Papers:
Some of the interesting machine learning papers published this week.
- Cut-and-Paste Neural Rendering
- Human-centric Dialog Training via Offline Reinforcement Learning
- Enhancing Extractive Text Summarization with Topic-Aware Graph Neural Networks
- Spherical Convolutional Neural Networks: Stability to Perturbations in SO(3)
- Statistical Inference for Online Decision-Making: In a Contextual Bandit Setting
- LTN: Long-Term Network for Long-Term Motion Prediction
- Auto Seg-Loss: Searching Metric Surrogates for Semantic Segmentation
- Compositional Embeddings for Multi-Label One-Shot Learning
- Uncertainty Aware Wildfire Management
- Anomaly Detection with Deep Perceptual Autoencoders