Given only a few frames of a video, humans can usually surmise what is happening and will happen on screen. If we see an early frame of stacked cans, a middle frame with a finger at the stack’s base, and a late frame showing the cans toppled over, we can guess that the finger knocked down the cans. Computers, however, struggle with this concept.

In a paper presented at the European Conference on Computer Vision, MIT researchers describe an add-on module that helps artificial intelligence systems called convolutional neural networks, or CNNs, to fill in the gaps between video frames to greatly improve the network’s activity recognition

IIT Bombay student develops a device to locate veins before drawing blood.

For many of us, the thought of being pricked by a needle to draw blood or inject drugs is horrifying, right? What if you had to be pierced many times because the right vein could not be identified? Nightmarish you say? Soon, this could be the thing of past, thanks to an award-winning ‘vein tracer’ by Mr. Trivikram Annamali, a student of the Indian Institute of Technology Bombay.

The vein tracer is a simple, hand-held, lightweight device that can help medical practitioners identify the veins during a venipuncture procedure like drawing blood or administering an intravenous injection. It works using near-infrared spectroscopy, where infrared light is used to understand the interaction between matter (blood, in this case) and the infrared radiation.

MIT computer scientists have developed a system that learns to identify objects within an image, based on a spoken description of the image. Given an image and an audio caption, the model will highlight in real-time the relevant regions of the image being described.

Unlike current speech-recognition technologies, the model doesn’t require manual transcriptions and annotations of the examples it’s trained on. Instead, it learns words directly from recorded speech clips and objects in raw images, and associates them with one another.

The model can currently recognize only several hundred different words and object types. But the researchers hope that one day their combined speech-object recognition technique could save countless hours of manual labor and open new doors in speech and image recognition.

For one thing, once you pick up the pen, you then have to shift it in your hand to get it in the writing position. “You move the pen from an initial grasp to something more stable, but to do that, you’re gripping it with certain surfaces of the fingers and sliding it against other surfaces of the fingers,” said Adam Spiers, a former associate research scientist in Aaron Dollar’s lab. “It’s really quite complicated as far as all the force and frictional interactions go, but we all take it for granted as a very simple thing to do.”

So Dollar’s lab tried a simpler tack that focuses on the variation in friction of finger surfaces. Their results were recently published in IEEE Robotics and Automation Letters and will also soon be presented at the International Conference on Intelligent Robots and Systems (IROS 2018).

Although it emerged from the field of cryptocurrency, blockchain’s tamper-proof, decentralized technology has important implications not only for the financial industry but for real estate, insurance, the law and other areas. Blockchain functions like a constantly-updated database distributed among a network of computers, instead of being controlled by a single entity.

Emin Gün Sirer, associate professor of computer science and co-director of Cornell’s Initiative for CryptoCurrencies and Contracts (IC3), says the widespread adoption of blockchains and smart contracts is going to be an “extinction-level event” for many companies.