What are the hottest research trends in artificial intelligence?
Below are the eleven areas highlighted in a new Standard University report on artificial intelligence:
1) Large-scale machine learning: Researchers are trying to scale existing machine-learning algorithms so they’ll work on very large data sets.
2) Deep learning: This branch of machine learning tries to mimic the activity in the layers of neurons in the human neocortex. It’s making progress in areas such as object recognition, video labeling, activity recognition as well as audio, speech, and natural language processing.
3) Reinforcement learning: It focuses to decision making to help AI systems learn about and execute actions in the real world. The recent success of AlphaGo, a computer program that beat the human Go champion, was due in large part to reinforcement learning.
4) Robotics: The main goal today is to train a robot to interact with the world around it in generalizable and predictable ways.
5) Computer vision: This is the most prominent form of machine perception and has made great progress thanks to a combination of large-scale computing, the availability of huge datasets on the Internet, and refinements to neural network algorithms. Today, computers can perform some visual tasks better than people.
6) Natural Language Processing: This is coupled with automatic speech recognition and has made huge practical progress in the marketplace. Google announced that 20% of current mobile queries are done by voice.
7) Collaborative systems: The goal is to develop autonomous systems that can work collaboratively with other systems and with humans. There’s a growing interest in applications that can leverage the complementary strengths of humans and machines.
8) Crowdsourcing and human computation: The focus is on augmenting computer systems by using crowdsourcing and human intelligence to solve problems. “Citizen science platforms energize volunteers to solve scientific problems, while paid crowdsourcing platforms such as Amazon Mechanical Turk provide automated access to human intelligence on demand,” notes the report.
9) Algorithmic game theory and computational social choice: This field has, for example, allowed game theorists to develop a virtually unbeatable Poker-playing algorithm.
10) Internet of Things (IoT): AI can process the huge amounts of data being output by Internet-connected devices such as appliances, vehicles, buildings, and cameras. Since devices often use incompatible communication protocols, AI might be able to help translate and make sense of this Tower of Babel.
11) Neuromorphic Computing: The idea is to build a new computing model to replace or amend the traditional von Neumann model of computing (that is, the computers we all currently use). Neuromorphic computers could mimic aspects of biological neural networks to improve hardware efficiency and robustness. The Standford report notes, “A larger wave may hit when these networks can be trained and executed on dedicated neuromorphic hardware, as opposed to simulated on standard von Neumann architectures, as they are today.”