GROOT is a new imitation learning technique developed by researchers at The University of Texas at Austin and Sony AI. It addresses the challenge of enabling robots to perform well in real-world settings with changing backgrounds, camera viewpoints, and object instances. GROOT focuses on building object-centric 3D representations and uses a transformer-based strategy to reason…

MLCommons has formed the AI Safety Working Group (AIS) to develop benchmarks for AI safety. Currently, there is no standardized benchmark to compare the safety of different AI models. AIS will build upon the Holistic Evaluation of Language Models (HELM) framework developed by Stanford University to create safety benchmarks for large language models. Several prominent…

AutoMix is an innovative approach to allocating queries to language models (LLMs) based on the correctness of responses. It uses context and self-verification to ensure accuracy, and can switch between different models. AutoMix enhances performance and computational cost in language processing tasks and demonstrates promising capabilities for future research and application.

NYU researchers have developed an “interpretable-by-design” machine learning model for understanding RNA splicing. While traditional machine learning models struggle with interpretability, this model not only provides accurate predictions but also explains the underlying biological processes. It achieves this by utilizing sequence and structure filters, assigning quantitative strengths to these filters, and introducing visualization tools. This…

A research team has developed a comprehensive set of metrics to evaluate the performance of deep generative models (DGMs) in engineering design. These metrics address aspects such as design constraints, diversity, novelty, and target achievement, providing a more holistic understanding of the capabilities and limitations of DGMs. The integration of these metrics allows for the…

Global payment leader Mastercard has partnered with crypto payment platform MoonPay to leverage Web3 tools for improved marketing and customer engagement. The collaboration was announced at the Money20/20 event in Las Vegas, with both companies expressing enthusiasm for creating enhanced experiences using Web3. Mastercard has been actively exploring Web3 initiatives and previously collaborated with companies…

Recent advancements in human action recognition have facilitated significant breakthroughs in Human-Robot Interaction (HRI). To achieve better action segmentation models, a team of researchers proposed a novel learning technique that maximizes the likelihood of action union for unlabeled frames. They also introduced a refining method during inference to enhance the accuracy of action labels. These…

Researchers have developed an algorithm called EUREKA that uses advanced LLMs, such as GPT-4, to create reward functions for complex skill acquisition through reinforcement learning. EUREKA outperforms human-engineered rewards and enables in-context learning based on human feedback. This breakthrough opens up possibilities for LLM-powered skill acquisition, as demonstrated by a simulated Shadow Hand mastering pen…

Google has invested $2 billion in Anthropic, an AI startup, making it a major contender in the industry alongside established players like OpenAI. The funding deal includes an immediate $500 million, with a potential commitment of up to $1.5 billion later. Anthropic aims to challenge OpenAI with its enterprise-focused approach and its development of a…

The rise of deep fakes poses a significant challenge for the AI industry. In 2023, there has been an influx of deep fake images and voice recordings, including fake news related to the Israel-Hamas conflict. The prevalence of AI-generated fakes has led to doubts about the authenticity of real content. The issue lies in the…