Welcome to our monthly digest, where you can catch up on all the AIhub stories you might have missed, browse the latest news, recap recent events, and much more. This month, we take a journey to RoboCup 2024, explore the International Conference on Machine Learning, and discover interactive real-world simulators.
Robots Galore at RoboCup 2024
RoboCup is an international scientific initiative aimed at advancing the state-of-the-art in intelligent robots. Throughout the year, a series of competitions and meetings are organized under this initiative. The flagship event is an international affair, with teams from around the world testing their machines.
This year, RoboCup 2024 took place in Eindhoven, attracting around 2,000 roboticists. Although RoboCup started as a soccer competition, other leagues have since been introduced, focusing on robots in industrial, rescue, and domestic environments. There is even a league dedicated to young roboticists.
We had the opportunity to attend in person this year and provided daily summaries of the action, focusing on different leagues each day. You can read the articles here: July 19, July 20, and July 21.
Interview with Sherry Yang: Learning from Interactive Real-World Simulators
Sherry Yang, Yilun Du, Kamyar Ghasemipour, Jonathan Tompson, Leslie Kaelbling, Dale Schuurmans, and Pieter Abbeel won an Outstanding Paper Award at ICLR2024 for their work on Learning from Interactive Real-World Simulators. Their paper presents a universal simulator (called UniSim) that uses images and text to train a robot simulator. We spoke to Sherry about this work, the challenges faced, and potential applications.
Are Models Biased on Text Without Gender-Related Language?
In their work, Are Models Biased on Text Without Gender-Related Language?, Catarina G Belém, Preethi Seshadri, Yasaman Razeghi, and Sameer Singh audit 28 popular language models to determine if they exhibit gender biases in stereotype-free contexts. In this blog post, Catarina summarizes their findings, noting that their results suggest gender biases do not solely stem from the presence of gender-related words in sentences.
Interview with Yuan Yang: Working at the Intersection of AI and Cognitive Science
You may have seen our interview series where we meet participants of the AAAI/SIGAI Doctoral Consortium to learn more about their research. In this latest interview, we hear from Yuan Yang, who completed his PhD in May. Yuan’s work lies at the intersection of AI and cognitive science, focusing on how AI can help understand fundamental human cognitive abilities and how such understanding can facilitate AI development.
International Conference on Machine Learning
Another major event this month was the International Conference on Machine Learning (ICML2024). Taking place in Vienna, the six main conferences covered topics ranging from open science to youth development, African languages to particle physics. You can see what participants were up to in our social media roundup. During the conference, the Test of Time and Best Paper awards were announced. You can see the winners here.
Summarizing Three ICML Papers
Speaking of ICML, Prabhu Prakash Kagitha has written accessible summaries of three papers he found particularly interesting: "Generalizing from Weak to Strong: Achieving Strong Capabilities with Weak Supervision", "Interpreting and Improving Large Language Models in Arithmetic Reasoning", and "Monitoring AI-Modified Content at Scale: A Case Study on the Impact of ChatGPT on AI Conference Peer Reviews". You can read them here.
Participating in the Mathematics Olympiad
DeepMind’s AlphaProof and AlphaGeometry 2 models were used to tackle this year’s International Mathematical Olympiad, which includes six advanced reasoning problems. DeepMind reports that the systems solved four out of six problems, reaching the same level as a silver medalist in the competition. Mathematician Timothy Gowers provides context and nuances the result in this informative thread.
Building a Better Future with Data and AI
On July 24, the Open Data Institute (ODI) released a white paper titled: “Building a Better Future with Data and AI”. The paper outlines ODI’s vision for AI in the UK, emphasizing the need for robust data infrastructure, governance, and ethical foundations to support the tech ecosystem.
Machine Learning and Logistic Regression
IBM’s explainer video series continues as Diarra Bell explains the basics of logistic regression, its application in binary classification, and how it can be used to predict probabilities using the sigmoid function.
—
Our resources page
Our events page
Seminars in 2024
AAAI/ACM SIGAI Doctoral Consortium Interview Series
AI Around the World Series
New Voices in AI Series
Keywords: AAAI, AAAI2024, ICML, ICML2024, monthly digest, RoboCup, RoboCup2024
—
Lucy Smith, Editor-in-Chief of AIhub
