This week’s list of top data news highlights covers August 10, 2024 to August 16, 2024, and includes articles on fine-tuning large language models (LLMs) and exploring their unexpected behavior and improving wheelchair racing with wearable technology.
Computer scientists at The Jackson Laboratory for Genomic Medicine in Connecticut have created a computer vision method to measure mouse body mass without handling the animals. Traditional methods required removing mice from their cages and placing them on scales, causing stress that could skew experimental results. This new approach uses video analysis to monitor the mice in their natural environment, reducing stress and improving the accuracy of research findings.
Genius Sports, a UK-based sports data company, has partnered with the English Premier League and is introducing a new offside-detection system called Dragon, powered by iPhones, which will replace the existing VAR system later in the 2024–25 season. This system uses multiple iPhone cameras to capture high-frame-rate video from various angles around the soccer field, allowing for more accurate and less controversial offside calls by tracking up to 10,000 data points on each player.
Clearly, a UK-based climate tech startup, has developed an AI-driven platform aimed at helping transport fleet operators reduce their carbon emissions. The platform consolidates anonymized data from logistics operators, including vehicle performance, GPS tracking, and load weight, to provide insights and alerts that can optimize driving behavior and improve fuel efficiency. The platform has already tracked over 100 million trips.
4. Delivering Goods Autonomously
Vayu Robotics, a California-based robotics startup, has launched an AI-powered delivery robot designed to enhance last-mile delivery services. The robot combines advanced AI with low-cost sensing technology to autonomously navigate streets, load orders in stores, and deliver packages directly to doorsteps. Vayu Robotics has already secured a contract to deploy 2,500 units in San Ramon, California.
Scientists from the University of Washington have used an machine learning tool to anaylze data from millions of galaxies, and have discovered that galaxies in denser areas of the universe tend to be larger those in less crowded, more isolated parts of the universe. Using data from the Subaru Telescope in Hawaii, the team analyzed highly detailed images of millions of galaxies of around 3 million galaxies in incredible detail.
Brian Seaman, a U.S. Paralympian in wheelchair racing, has been utilizing advanced wearable technology to enhance his racing performance for the Paris Games. An innovative sleeve, developed by the startup Nextiles, incorporates biometric threads that gather data from his elbow, which is then transmitted to a smartphone app to analyze his moves.
Methodist Mansfield Hospipital in Texas is using a tool from brain imaging analysis software company RAPID-AI to quickly identify potential abnormalities in scans. By processing vast amounts of imaging data from computed tomography (CT) scans, the software can swiftly identify abnormalities and notify medical staff, allowing for more rapid interventions.
Roto, a UK-based virtual reality (VR) entertainment company, has developed a chair that spins the user in the chair to match their head turns, which aligns physical movement with what the user sees in the VR headset. This aims to make VR feel more immersive and reduce motion sickness. The chair was made in partnership with Meta and works with its Quest VR headsets that have patented “Look & Turn” technology.
Lambda, an California-based AI infrastructure company, has partnered with Nous Research, an applied research group focused on LLMs, to create Hermes 3, a fine-tuned version of Meta’s Llama 3.1 model. Available in three sizes,, Hermes 3 stands out for its ability to retain long-term context and manage multiturn conversations. Despite its strengths, the model produced unexpected outputs when given a blank prompt, including responses that appeared to reflect uncertainty or confusion, which was not anticipated by its developers. The creators are inviting users to explore further.
Airbus is leveraging an AI platform from U.S.-based geospatial company Atlas AI to analyze travel demand, airport catchment areas, and infrastructure developments. Atlas AI leverages satellite imagery, demographic data, land use patterns, and other metrics to precisely forecast market dynamics and predict future demand.
Image credits: Wikimedia commons