LLM-generated plans in abstract spaces lack physical grounding.
Human demonstrations in high-dimensional continuous spaces lack labels for motion constraints.
MIT CSAIL-led research grounds language plans in demonstrations as “manipulation modes” to build robust policies: bit.ly/3Q3YcGj
A thread 🧵
Great #ML research requires great systems. Here we discuss some strategies we’re using to help serve and train sophisticated ML models while easing the complexity of implementation for end users. Read more at goo.gle/3JEbmr6
We’re hitting a tipping point for #ArtificialIntelligence: With #ChatGPT and other AI models that can communicate in plain English, write and revise text, and write code, the technology is suddenly becoming more useful to a broader population of people.
hbr.org/2022/12/chatgp…
Understanding how neural networks reach decisions can be challenging, but is also valuable for uses from image analysis to scientific discovery. A new approach, called StylEx, discovers and visualizes how disentangled attributes affect a classifier → goo.gle/357gn9r
Clustering algorithms partition datasets into meaningful groups and are a key building block of unsupervised #ML. Learn about a new clustering algorithm that enhances privacy while maintaining or improving performance against existing benchmarks ↓ goo.gle/3m2sxWO
Identifying whether a chest X-ray exhibits abnormalities is critical for patient triage and care. Today we demonstrate a method that uses #ML to rapidly assess and flag potentially abnormal chest X-rays for radiologist review. Read all about it below. goo.gle/3yGZdIQ
Last year we partnered with @NSF to support a National AI Research Institute focused on Human-AI Interaction and Collaboration. Today, they have selected the AI-CARING institute, led by @GeorgiaTech & others, to receive a $20M grant. Learn more below! blog.google/technology/ai/…