Augmented reality NBA analysis; information extraction; pectin memristors.
Revolutionizing sports via AI and computer vision
A new technology developed by PlayfulVision — an EPFL startup — will be used in all NBA games in the United States starting next year to records all aspects of sporting events for subsequent analysis in augmented reality.
Will artificial intelligence and computer vision revolutionize the sports industry? PlayfulVision’s approach uses a multi-camera system to capture and analyze the athletes’ every move with the aim is to see how players pull off a great pass, for example, and to assess whether or not certain shots should have been taken. This technology was developed at EPFL and then bought by Second Spectrum last year. It caught the attention of the largest sports leagues in the world, including the NBA, which has just sealed a seven-year deal to use it in all official games starting in 2017.
The system is based on optical trackers that capture every move by every player and by the ball, recording their exact coordinates 25 times per second. This information will be used to carefully analyze games: body position, shots, rebounds and fouls — to provide an in-depth understanding of the games, at a greater level of detail than a coach can perceive alone.
AI system surfs web to improve performance
Of the vast wealth of information unlocked by the Internet, most is plain text, and extracting the data necessary and organizing it for quantitative analysis to answer myriad questions, about, for instance, the correlations between the industrial use of certain chemicals and incidents of disease, or between patterns of news coverage and voter-poll results — may be prohibitively time consuming, according to MIT researchers.
As such, information extraction — or automatically classifying data items stored as plain text — is a major topic of artificial-intelligence research. And recently, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory won a best-paper award for a new approach to information extraction that turns conventional machine learning on its head.
Most machine-learning systems work by combing through training examples and looking for patterns that correspond to classifications provided by human annotators, the researchers reminded. For instance, humans might label parts of speech in a set of texts, and the machine-learning system will try to identify patterns that resolve ambiguities — for instance, when “her” is a direct object and when it’s an adjective. And typically, computer scientists will try to feed their machine-learning systems as much training data as possible to increase the chances that a system will be able to handle difficult problems.
However, in their new paper, the MIT team trained their system on scanty data — because in the scenario they’re investigating, that’s usually all that’s available and then , interestingly, they find the limited information an easy problem to solve.
Advancing neuron-like electronic systems with pectin
While most of us know pectin as a key ingredient for making jam and jelly, a team of University of Parma researchers know it as a component for a complex hybrid device that links biological and electronic systems.
The team built on previous work in this field using pectin with a high degree of methylation as the medium to create a new architecture of hybrid device with a double-layered polyelectrolyte that alone drives memristive behavior. A memristive device can be thought of as a synapse analogue, a device that has a memory. Simply stated, its behavior in a certain moment depends on its previous activity, similar to the way information in the human brain is transmitted from one neuron to another.
To create the hybrid device, the researchers applied materials generally used in the pharmaceutical and food industries in the electrochemical devices. The idea of using the ‘buffering’ capability of these biocompatible materials as solid polyelectrolyte is completely innovative and this is believed to be the first time that these bio-polymers have been used in devices based on organic polymers and in a memristive device.
Memristors can provide a bridge for interfacing electronic circuits with nervous systems, moving us closer to realization of a double-layer perceptron, an element that can perform classification functions after an appropriate learning procedure, the team noted.