Technology and Science Highlights of the Week #24 September 2024
Google, in cooperation with the Indian startup Salcit Technologies, is working on artificial intelligence that can detect early signs of diseases based on sounds. The company trains its basic AI model on 300 million audio recordings, including coughs, snorts, and heavy breathing.
Salcit Technologies is engaged in the creation of artificial intelligence for respiratory diseases
This is not Google's first attempt to digitize the human senses
Last week, the automated window cleaning system
“Tianmushan No. 1,” the world’s first hydrogen-powered multi-rotor drone with a range of 100 km, developed by China’s Tianmushan Laboratory, has completed its maiden flight.
“The drone was developed to solve the industry’s problems related to the insufficient endurance and resistance of industrial drones to low temperatures,” the laboratory said in a press release.
The main feature of the drone is its hydrogen fuel cell
BEIJING, Aug 26 (Reuters) - As China seeks to race ahead in humanoid robot development, its supply chains showcased cheaper and innovative parts at the World Robot Conference in Beijing, but some executives warn the industry has yet to improve product reliability.
Wisson Technology (Shenzhen), known for its flexible robotic manipulators, doesn’t depend on motors and reducers - transmission devices commonly used in robotics - but instead uses 3D-printed plastics and relies on pneumatic artificial muscles to power its robots.
Recommended by LinkedIn
"(Wisson’s) pliable arms could be used in humanoids," said Cao, adding that the company has already provided samples to overseas companies that make humanoid robots, without elaborating.
In their paper, a team of researchers from the Alberta Machine Intelligence Institute (Amii) explores a difficult problem that has long been suspected in deep learning models but hasn’t received much attention: For some reason, many deep learning agents engaged in continuous training lose their ability to learn, and their performance drops off dramatically.
Many existing models aren’t designed to learn continuously. Sutton cites ChatGPT as an example; it doesn’t learn continuously. Instead, its creators train the model for a set amount of time. Once training is complete, the model is deployed without further training.
The researchers proposed a method called “continuous backpropagation” that helps models retain their learning ability for longer, solving this problem. This discovery could be an important step toward creating more robust and efficient AI systems
Geophysicists from Australia have discovered an unusual structure in the liquid core of the Earth, which may improve our understanding of the dynamics of the planet's magnetic field.
The Earth has a complex internal structure, which consists of several layers. In the center of the planet is the core, which is divided into two parts: a solid inner and a liquid outer. Above the core lies the mantle, and the outer layer is formed by the Earth's crust. Recently, a team of researchers from the Australian National University made an important discovery. They found a previously unknown structure in the outer core of the Earth, near the boundary with the mantle.
Professor Hrvoje Tkalčić, co-author of the study, explains that the new structure was discovered thanks to the slowing down of seismic waves. According to him, it resembles a donut shape and extends for the first hundreds of kilometers below the boundary of the mantle and the core.
The outer core of the Earth consists mainly of molten iron and nickel. The movements of this electrically conductive mass create a magnetic field that protects the planet from cosmic radiation. Professor Tkalčić suggests that the discovered structure is enriched with light chemical elements, which enhances mixing in the outer core.
The geophysicist also notes that this discovery may be important not only for studying the Earth. New knowledge about the structure of the Earth's core and its influence on the planet's magnetic field may be useful for studying the magnetic fields of other planets.
See You Next Week with News!