Latest News Machine Learning Tech news
Particle physicists have taught algorithms to solve previously unsolvable issues.
Hey everyone! I upgraded a previous redstone build to support 3D Wireframe Rendering! Thanks everyone who suggested this, it was a lot of fun! bigsmile
!!! WATCH PART 1 HERE!!!
https://youtu.be/vfPGuUDuwmo.
Continue reading “I Made a 3D Renderer with just redstone!” »
If you’ve ever played the claw game at an arcade, you know how hard it is to grab and hold onto objects using robotics grippers. Imagine how much more nerve-wracking that game would be if, instead of plush stuffed animals, you were trying to grab a fragile piece of endangered coral or a priceless artifact from a sunken ship.
Most of today’s robotic grippers rely on embedded sensors, complex feedback loops, or advanced machine learning algorithms, combined with the skill of the operator, to grasp fragile or irregularly shaped objects. But researchers from the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS) have demonstrated an easier way.
Continue reading “Tentacle robot can gently grasp fragile objects” »
Researchers can now predict exactly how soap molecules spread across a body of water, an everyday but surprisingly complex process.
When a tiny drop of soapy water falls onto a pool of liquid, its contents spread out over the pool’s surface. The dynamics of this spreading depend on the local concentration of soap—which varies in time and is difficult to predict—at each point across the entire pool’s surface. Now Thomas Bickel of the University of Bordeaux in Talence, France, and Francois Detcheverry of the University of Lyon, France, have derived an exact time-dependent solution for these distributions [1]. The solution reveals surprisingly rich behaviors in this everyday phenomenon.
The duo considered a surfactant-laden drop spreading over the surface of a deep pool of fluid. Researchers have previously shown that the equations governing the transport of the surfactant particles can be mapped to a partial differential equation known as the Burgers’ equation, which was initially developed to describe flows in turbulent fluids.
A Brazilian study published in Scientific Reports shows that artificial intelligence (AI) can be used to create efficient models for genomic selection of sugarcane and forage grass varieties and predict their performance in the field on the basis of their DNA.
In terms of accuracy compared with traditional breeding techniques, the proposed methodology improved predictive power by more than 50%. This is the first time a highly efficient genomic selection method based on machine learning has been proposed for polyploid plants (in which cells have more than two complete sets of chromosomes), including the grasses studied.
Machine learning is a branch of AI and computer science involving statistics and optimization, with countless applications. Its main goal is to create algorithms that automatically extract patterns from datasets. It can be used to predict the performance of a plant, including whether it will be resistant to or tolerant of biotic stresses such as pests and diseases caused by insects, nematodes, fungi or bacteria, and or abiotic stresses such as cold, drought, salinity or insufficient soil nutrients.
In quantum physics, Fermi’s golden rule, also known as the golden rule of time-dependent perturbation theory, is a formula that can be used to calculate the rate at which an initial quantum state transitions into a final state, which is composed of a continuum of states (a so-called “bath”). This valuable equation has been applied to numerous physics problems, particularly those for which it is important to consider how systems respond to imposed perturbations and settle into stationary states over time.
Fermi’s golden rule specifically applies to instances in which an initial quantum state is weakly coupled to a continuum of other final states, which overlap its energy. Researchers at the Centro Brasileiro de Pesquisas Físicas, Princeton University, and Universität zu Köln have recently set out to investigate what happens when a quantum state is instead coupled to a set of discrete final states with a nonzero mean level spacing, as observed in recent many-body physics studies.
“The decay of a quantum state into some continuum of final states (i.e., a ‘bath’) is commonly associated with incoherent decay processes, as described by Fermi’s golden rule,” Tobias Micklitz, one of the researchers who carried out the study, told Phys.org. “A standard example for this is an excited atom emitting a photon into an infinite vacuum. Current date experimentations, on the other hand, routinely realize composite systems involving quantum states coupled to effectively finite size reservoirs that are composed of discrete sets of final states, rather than a continuum.”
Given the rapid pace at which technology is developing, it comes as no surprise that quantum technologies will become commonplace within decades. A big part of ushering in this new age of quantum computing requires a new understanding of both classical and quantum information and how the two can be related to each other.
Before one can send classical information across quantum channels, it needs to be encoded first. This encoding is done by means of quantum ensembles. A quantum ensemble refers to a set of quantum states, each with its own probability. To accurately receive the transmitted information, the receiver has to repeatedly ‘guess’ the state of the information being sent. This constitutes a cost function that is called ‘guesswork.’ Guesswork refers to the average number of guesses required to correctly guess the state.
The concept of guesswork has been studied at length in classical ensembles, but the subject is still new for quantum ensembles. Recently, a research team from Japan—consisting of Prof. Takeshi Koshiba of Waseda University, Michele Dall’Arno from Waseda University and Kyoto University, and Prof. Francesco Buscemi from Nagoya University—has derived analytical solutions to the guesswork problem subject to a finite set of conditions. “The guesswork problem is fundamental in many scientific areas in which machine learning techniques or artificial intelligence are used. Our results trailblaze an algorithmic aspect of the guesswork problem,” says Koshiba. Their findings are published in IEEE Transactions on Information Theory.
Quantum physics is strange. At least, it is strange to us, because the rules of the quantum world, which govern the way the world works at the level of atoms and subatomic particles (the behavior of light and matter, as the renowned physicist Richard Feynman put it), are not the rules that we are familiar with — the rules of what we call “common sense.”
The quantum rules, which were mostly established by the end of the 1920s, seem to be telling us that a cat can be both alive and dead at the same time, while a particle can be in two places at once. But to the great distress of many physicists, let alone ordinary mortals, nobody (then or since) has been able to come up with a common-sense explanation of what is going on. More thoughtful physicists have sought solace in other ways, to be sure, namely coming up with a variety of more or less desperate remedies to “explain” what is going on in the quantum world.
These remedies, the quanta of solace, are called “interpretations.” At the level of the equations, none of these interpretations is better than any other, although the interpreters and their followers will each tell you that their own favored interpretation is the one true faith, and all those who follow other faiths are heretics. On the other hand, none of the interpretations is worse than any of the others, mathematically speaking. Most probably, this means that we are missing something. One day, a glorious new description of the world may be discovered that makes all the same predictions as present-day quantum theory, but also makes sense. Well, at least we can hope.
Imagine taking a star twice the mass of the sun and crushing it to the size of Manhattan. The result would be a neutron star—one of the densest objects found anywhere in the universe, exceeding the density of any material found naturally on Earth by a factor of tens of trillions. Neutron stars are extraordinary astrophysical objects in their own right, but their extreme densities might also allow them to function as laboratories for studying fundamental questions of nuclear physics, under conditions that could never be reproduced on Earth.
Because of these exotic conditions, scientists still do not understand what exactly neutron stars themselves are made from, their so-called “equation of state” (EoS). Determining this is a major goal of modern astrophysics research. A new piece of the puzzle, constraining the range of possibilities, has been discovered by a pair of scholars at IAS: Carolyn Raithel, John N. Bahcall Fellow in the School of Natural Sciences; and Elias Most, Member in the School and John A. Wheeler Fellow at Princeton University. Their work was recently published in The Astrophysical Journal Letters.
Continue reading “New tool allows scientists to peer inside neutron stars” »
A while ago I spotted someone working on real time AI image generation in VR and I had to bring it to your attention because frankly, I cannot express how majestic it is to watch AI-modulated AR shifting the world before us into glorious, emergent dreamscapes.
Applying AI to augmented or virtual reality isn’t a novel concept, but there have been certain limitations in applying it—computing power being one of the major barriers to its practical usage. Stable Diffusion image generation software, however, is a boiled-down algorithm for use on consumer-level hardware and has been released on a Creative ML OpenRAIL-M licence. That means not only can developers use the tech to create and launch programs without renting huge amounts of server silicon, but they’re also free to profit from their creations.
