Machine learning could cut delays from traffic lights

Researchers have designed a new system that uses machine learning to improve the flow of vehicles at traffic lights.

---

machine-learning traffic science-and-technology

How explainable artificial intelligence can help humans innovate

AI algorithms can solve hard problems and learn incredible tasks, but they can't explain how they do these things. If researchers can build explainable AI, it could lead to a flood of new knowledge.

---

machine-learning innovation artificial-intelligence computer-science innovation-and-invention process-of-innovation black-box reinforcement-learning

Thousands of ocean fishing boats could be using forced labor – we used AI and satellite data to find them

Forced labor is a widespread problem in fisheries on the high seas. Between 2012 and 2018, an estimated 100,000 people may have been victims of forced labor on thousands of different boats.

---

machine-learning artificial-intelligence fisheries fishing human-rights satellite-data slavery forced-labor human-rights-violations

Cambridge launches Regulatory Genome Project

The project will use machine learning to sequence the world’s regulatory text and create an open-source repository of machine-readable regulatory information. 

---

machine-learning finance artificial-intelligence regulation alternative-finance

Making smart thermostats more efficient

A smart thermostat quickly learns to optimize building microclimates for both energy consumption and user preference.

---

machine-learning internet-of-things climate-change data research school-of-engineering electrical-engineering-computer-science-eecs algorithms idss laboratory-for-information-and-decision-systems-lids analytics mit-schwarzman-college-of-computing skoltech

Building machines that better understand human goals

A new algorithm capable of inferring goals and plans could help machines better adapt to the imperfect nature of human planning.

---

machine-learning artificial-intelligence research school-of-engineering school-of-science electrical-engineering-computer-science-eecs computer-science-and-technology algorithms brain-and-cognitive-sciences computer-science-and-artificial-intelligence-laboratory-csail mit-schwarzman-college-of-computing

It takes a lot of energy for machines to learn – here's why AI is so power-hungry

Training neural networks burns through a lot of energy. As the AI field grows, it's working to keep its carbon foot print from growing with it.

---

machine-learning computing climate-change artificial-intelligence energy carbon-emissions global-warming energy-efficiency neural-networks reducing-carbon-emissions data-centers

Model could help determine quarantine measures needed to reduce Covid-19’s spread

A key finding: Early reopening last spring led to a dramatic drop in “quarantine strength” in southern and west-central U.S. states.

---

machine-learning mathematics artificial-intelligence covid-19 pandemic disease research mechanical-engineering school-of-engineering school-of-science technology-and-society disaster-response

Neuroscientists find a way to make object-recognition models perform better

Adding a module that mimics part of the brain can prevent common errors made by computer vision models.

---

machine-learning artificial-intelligence research school-of-science brain-and-cognitive-sciences center-for-brains-minds-and-machines quest-for-intelligence mcgovern-institute defense-advanced-research-projects-agency-darpa mit-schwarzman-college-of-computing mit-ibm-watson-ai-lab