What's the news?: Deep learning is behind machine learning's most high-profile successes. But this incredible performance comes at a cost: training deep-learning models requires huge amounts of energy. Now, new research shows how scientists who use cloud platforms to train algorithms can dramatically reduce the energy they use, and therefore the emissions they create. How can they do it?: Simple changes to cloud settings are the key. Researchers created a tool that measures the electricity usage of any machine-learning program that runs on Azure, Microsoft's cloud service, during every phase of their project.

Since the first paper studying this technology's impact on the environment was published three years ago, a movement has grown among researchers to self-report the energy consumed and emissions generated from their work. Having accurate numbers is an important step toward making changes, but actually gathering those numbers can be a challenge. "You can't improve what you can't measure," says Jesse Dodge, a research scientist at the Allen Institute for AI in Seattle. "The first step for us, if we want to make progress on reducing emissions, is we have to get a good measurement." To that end, the Allen Institute recently collaborated with Microsoft, the AI company Hugging Face, and three universities to create a tool that measures the electricity usage of any machine-learning program that runs on Azure, Microsoft's cloud service.

One of the many technologies included under the umbrella of artificial intelligence, machine learning is defined by Wikipedia as "a field of computer science that gives computers the ability to learn without being explicitly programmed." The technology, which is a core part of the data analytics technologies that power the modern data warehouse, features algorithms that can make predictions on their own about data and its insights without being hampered by strict guidelines and instructions. When used successfully, machine learning can help with infrastructure scalability, cost savings, and agility. For its part, artificial intelligence (AI) is the ability of machines to think like humans. It stems from the idea that "given enough data and compute power, machines will be able to think and learn using mathematical simulation of the human brain," said John Santaferraro, research director at Enterprise Management Associates (EMA).

SYDNEY – After applying his machine-learning programs to central bank policy statements to churn out trading calls, a hedge fund-backed political economy specialist is training his sights on corporate earnings announcements. Evan Schnidman, a 31-year-old who set up his own firm after a Harvard University Ph.D. dissertation that looked at the Federal Reserve's communications, is hoping the approach that lured $3.3 million in a fundraising round last December will work in the corporate sphere. St. Louis-based Prattle has until now focused on applying the artificial intelligence method known as natural-language processing to make assessments of Fed and other central bank policy statements. At a time when analysis is poised to get its own price tag, with the introduction of Europe's MiFID II regulations, research costs are an increasing focus for investment banks and asset managers. BlackRock Inc. has even moved to use robots to design funds.

SYDNEY – After applying his machine-learning programs to central bank policy statements to churn out trading calls, a hedge fund-backed political economy specialist is aiming his sights on corporate earnings announcements. Evan Schnidman, a 31-year-old who set up his own firm after a Harvard University Ph.D. dissertation that looked at the Federal Reserve's communications, is hoping the approach that lured $3.3 million in a fundraising round last December will work in the corporate sphere. St. Louis-based Prattle has until now focused on applying the artificial intelligence method known as natural-language processing to make assessments of Fed and other central bank policy statements. At a time when analysis is poised to get its own price tag, with the introduction of Europe's MiFID II regulations, research costs are an increasing focus for investment banks and asset managers. BlackRock Inc. has even moved to use robots to design funds.

Artificial intelligence is advanced enough to do some pretty complicated things: read lips, mimic sounds, analyze photographs of food, and even design beer. Unfortunately, even people who have plenty of coding knowledge might not know how to create the kind of algorithm that can perform these tasks. Google wants to bring the ability to harness artificial intelligence to more people, though, and according to WIRED, it's doing that by teaching machine-learning software to make more machine-learning software. The project is called AutoML, and it's designed to come up with better machine-learning software than humans can. As algorithms become more important in scientific research, healthcare, and other fields outside the direct scope of robotics and math, the number of people who could benefit from using AI has outstripped the number of people who actually know how to set up a useful machine-learning program.

Researchers have tapped into artificial intelligence to help overcome some of fusion energy's greatest challenges. By feeding a machine-learning program data from past experiments, it can reveal links between processes that cause complications in the plasma's behaviour This could help to avoid such disruptions, which lead to rapid loss of stored thermal and magnetic energy, and can even threaten the machine itself. According to the researchers, this approach could be used to analyze the behaviour of plasma inside a tokamak. Fusion involves placing hydrogen atoms under high heat and pressure until they fuse into helium atoms. When deuterium and tritium nuclei - which can be found in hydrogen - fuse, they form a helium nucleus, a neutron and a lot of energy.

In debates over the future of artificial intelligence, many experts think of these machine-based systems as coldly logical and objectively rational. But in a new study, Princeton University-based researchers have demonstrated how machines can be reflections of their creators in potentially problematic ways. Common machine-learning programs trained with ordinary human language available online can acquire the cultural biases embedded in the patterns of wording, the researchers reported in the journal Science April 14. These biases range from the morally neutral, such as a preference for flowers over insects, to discriminatory views on race and gender. Identifying and addressing possible biases in machine learning will be critically important as we increasingly turn to computers for processing the natural language humans use to communicate, as in online text searches, image categorization and automated translations.

Investigators eager to uncover the genetic basis of autism could now have hundreds of promising new leads thanks to a study by Princeton University and Simons Foundation researchers. In the first effort of its kind, the research team developed a machine-learning program that scoured the whole human genome to predict which genes may contribute to autism spectrum disorder (ASD). The results of the program's analyses -- a rogue's gallery of 2,500 candidate genes -- vastly expand on the 65 autism-risk genes currently known. Researchers have recently estimated that 400 to 1,000 genes underpin the complex neurodevelopmental disorder. This newest research provides a manageable, "highly enriched" pool from which to pin down the full suite of ASD-related genes, the researchers said.

Investigators eager to uncover the genetic basis of autism could now have hundreds of promising new leads thanks to a study by Princeton University and Simons Foundation researchers. In the first effort of its kind, the research team developed a machine-learning program that scoured the whole human genome to predict which genes may contribute to autism spectrum disorder (ASD). The results of the program's analyses -- a rogue's gallery of 2,500 candidate genes -- vastly expand on the 65 autism-risk genes currently known. Researchers have recently estimated that 400 to 1,000 genes underpin the complex neurodevelopmental disorder. This newest research provides a manageable, "highly enriched" pool from which to pin down the full suite of ASD-related genes, the researchers said.