The Next Domain that Machine Learning Aims to Conquer is Hard Science
Hard sciences are also being revolutionized by machine learning
From email to the Internet, particle physicists, the hard sciences’ experts, have historically been early users of technology, if not its creators. Therefore, it is not unexpected that researchers began training computer models to tag particles in the chaotic jets produced by collisions as early as 1997. Since then, these models have plodded along, becoming increasingly more capable—although not everyone has been pleased with this development. Particle physicists have taught algorithms to solve previously unsolvable issues and take on entirely new challenges over the past ten years, concurrently with the broader deep-learning revolution.
According to Jesse Thaler, a theoretical particle physicist at the Massachusetts Institute of Technology, “I felt really scared by machine learning.” He claims that at first, he believed that it imperiled his ability to characterize particle jets using human judgment. Thaler, however, has now come to accept it and has used machine learning to solve a number of issues in particle physics. He claims that machine learning is a partner.
To begin with, the data utilized in particle physics differs greatly from the conventional data used in machine learning. Though convolutional neural networks (CNNs) have excelled at categorizing photos of commonplace items like trees, kittens, and food, they’re less good at handling particle collisions. Javier Duarte, a particle physicist at the University of California, San Diego, claims that the issue is that collision data from sources like the Large Hadron Collider isn’t by nature an image. Flashy representations of LHC collisions may deceitfully fill the entire detector. In reality, a white screen with a few black pixels represents the millions of inputs that aren’t actually registering a signal. Although this weakly supplied data produces a subpar image, it can perform well in a newer architecture called graph neural networks (GNNs).
Innovation is needed to overcome additional particle physics problems. According to Daniel Whiteson, a particle physicist at the University of California, Irvine, “We’re not merely importing hammers to smash our nails.” We need to create new hammers because there are strange new types of nails. The enormous volume of data generated at the LHC—roughly one petabyte every second—is one peculiar nail. Only a limited amount of high-quality data is saved from this large volume. Researchers seek to teach a sharp-eyed algorithm to sort better than one that is hard coded in order to develop a better trigger system, which saves as much good data as possible while getting rid of low-quality data. The intention is not to connect the device or the experiment to the network and have it publish the articles without keeping them informed, according to Whiteson. He and his colleagues are attempting to have the algorithms deliver feedback in terms of what people can comprehend, but it’s possible that other individuals have communication duties as well.
However, according to Duarte, such an algorithm would need to execute in just a few microseconds in order to be efficient. Particle physicists are pushing the boundaries of machine techniques like pruning and quantization to accelerate their algorithms in order to solve these issues. Researchers are looking for ways to compress the data because the LHC needs to store 600 petabytes during the next five years of data collecting (equal to about 660,000 movies at 4K resolution or the data equivalent of 30 Libraries of Congresses).
As some particle physicists go deeper into machine learning it seems imperative to ask, “Are they doing physics or computer science?” This is a troubling topic. There is already a stigma surrounding coding, which is occasionally dismissed as “not real physics,” and similar worries abound over machine learning. One concern is that machine learning will muddle the physics, making analysis a black box of automated operations that are difficult for humans to comprehend. On the one hand, Thaler argues, “We’d like to have a machine learn to think more like a physicist, on the other hand, we only need to learn how to think a little bit more like a computer.” “We must learn to communicate in each other’s languages.
The post The Next Domain that Machine Learning Aims to Conquer is Hard Science appeared first on Analytics Insight.
Hard sciences are also being revolutionized by machine learning
From email to the Internet, particle physicists, the hard sciences’ experts, have historically been early users of technology, if not its creators. Therefore, it is not unexpected that researchers began training computer models to tag particles in the chaotic jets produced by collisions as early as 1997. Since then, these models have plodded along, becoming increasingly more capable—although not everyone has been pleased with this development. Particle physicists have taught algorithms to solve previously unsolvable issues and take on entirely new challenges over the past ten years, concurrently with the broader deep-learning revolution.
According to Jesse Thaler, a theoretical particle physicist at the Massachusetts Institute of Technology, “I felt really scared by machine learning.” He claims that at first, he believed that it imperiled his ability to characterize particle jets using human judgment. Thaler, however, has now come to accept it and has used machine learning to solve a number of issues in particle physics. He claims that machine learning is a partner.
To begin with, the data utilized in particle physics differs greatly from the conventional data used in machine learning. Though convolutional neural networks (CNNs) have excelled at categorizing photos of commonplace items like trees, kittens, and food, they’re less good at handling particle collisions. Javier Duarte, a particle physicist at the University of California, San Diego, claims that the issue is that collision data from sources like the Large Hadron Collider isn’t by nature an image. Flashy representations of LHC collisions may deceitfully fill the entire detector. In reality, a white screen with a few black pixels represents the millions of inputs that aren’t actually registering a signal. Although this weakly supplied data produces a subpar image, it can perform well in a newer architecture called graph neural networks (GNNs).
Innovation is needed to overcome additional particle physics problems. According to Daniel Whiteson, a particle physicist at the University of California, Irvine, “We’re not merely importing hammers to smash our nails.” We need to create new hammers because there are strange new types of nails. The enormous volume of data generated at the LHC—roughly one petabyte every second—is one peculiar nail. Only a limited amount of high-quality data is saved from this large volume. Researchers seek to teach a sharp-eyed algorithm to sort better than one that is hard coded in order to develop a better trigger system, which saves as much good data as possible while getting rid of low-quality data. The intention is not to connect the device or the experiment to the network and have it publish the articles without keeping them informed, according to Whiteson. He and his colleagues are attempting to have the algorithms deliver feedback in terms of what people can comprehend, but it’s possible that other individuals have communication duties as well.
However, according to Duarte, such an algorithm would need to execute in just a few microseconds in order to be efficient. Particle physicists are pushing the boundaries of machine techniques like pruning and quantization to accelerate their algorithms in order to solve these issues. Researchers are looking for ways to compress the data because the LHC needs to store 600 petabytes during the next five years of data collecting (equal to about 660,000 movies at 4K resolution or the data equivalent of 30 Libraries of Congresses).
As some particle physicists go deeper into machine learning it seems imperative to ask, “Are they doing physics or computer science?” This is a troubling topic. There is already a stigma surrounding coding, which is occasionally dismissed as “not real physics,” and similar worries abound over machine learning. One concern is that machine learning will muddle the physics, making analysis a black box of automated operations that are difficult for humans to comprehend. On the one hand, Thaler argues, “We’d like to have a machine learn to think more like a physicist, on the other hand, we only need to learn how to think a little bit more like a computer.” “We must learn to communicate in each other’s languages.
The post The Next Domain that Machine Learning Aims to Conquer is Hard Science appeared first on Analytics Insight.
Denial of responsibility! Techno Blender is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.