OpenAI’s latest neural network creates images from written descriptions

OpenAI has debuted its latest jaw-dropping innovation, an image-generating neural network called DALL·E.

DALL·E is a 12-billion parameter version of GPT-3 which is trained to generate images from text descriptions.

“We find that DALL·E is able to create plausible images for a great variety of sentences that explore the compositional structure of language,“ OpenAI explains.

Generated images can range from drawings, to objects, and even manipulated...

Researchers achieve 94% power reduction for on-device AI tasks

Researchers from Applied Brain Research (ABR) have achieved significantly reduced power consumption for a range of AI-powered devices.

ABR designed a new neural network called the Legendre Memory Unit (LMU). With LMU, on-device AI tasks – such as those on speech-enabled devices like wearables, smartphones, and smart speakers – can take up to 94 percent less power.

The reduction in power consumption achieved through LMU will be particularly beneficial to smaller...

Over 1,000 researchers sign letter opposing ‘crime predicting’ AI

More than 1,000 researchers, academics, and experts have signed an open letter opposing the use of AI to predict crime.

Anyone who has watched the sci-fi classic Minority Report will be concerned about attempts to predict crime before it happens. In an ideal scenario, crime prediction could help determine where to allocate police resources – but the reality will be very different.

The researchers are speaking out ahead of an imminent publication titled ‘A Deep Neural...

Nvidia comes out on top in first MLPerf inference benchmarks

nvidia mlperf benchmarks mlperf ai artificial intelligence neural networks ml machine learning

The first benchmark results from the MLPerf consortium have been released and Nvidia is a clear winner for inference performance.

For those unaware, inference takes a deep learning model and processes incoming data however it’s been trained to.

MLPerf is a consortium which aims to provide “fair and useful” standardised benchmarks for inference performance. MLPerf can be thought of as doing for inference what SPEC does for benchmarking CPUs and general system...

Intel unwraps its first chip for AI and calls it Spring Hill

intel spring hill ai artificial intelligence chip processor

Intel has unwrapped its first processor that is designed for artificial intelligence and is planned for use in data centres.

The new Nervana Neural Network Processor for Inference (NNP-I) processor has a more approachable codename of Spring Hill.

Spring Hill is a modified 10nm Ice Lake processor which sits on a PCB and slots into an M.2 port typically used for storage.

According to Intel, the use of a modified Ice Lake processor allows Spring Hill to handle...

EmoNet: Emotional neural network automatically categorises feelings

emonet neural network ai artificial intelligence emotions feelings

A neural network called EmoNet has been designed to automatically categorise the feelings of an individual.

EmoNet was created by researchers from the University of Colorado and Duke University and could one day help AIs to understand and react to human emotions.

The neural network is capable of accurately classifying images into 11 emotions, although some with a higher confidence than others.

‘Craving,’ ‘sexual desire,’ and ‘horror’ were able to be...

Samsung researchers create AI which generates realistic 3D renders of video scenes

Samsung researchers create AI which generates realistic 3D renders of video scenes

Three researchers at Samsung have created an AI which can generate realistic 3D renders of video scenes.

In a paper detailing the neural network behind the AI, the researchers explained the inefficient process of creating virtual scenes today:

“Creating virtual models of real scenes usually involves a lengthy pipeline of operations. Such modeling usually starts with a scanning...

MIT’s AI uses wireless signals to detect movement through walls

mit csail wall rfpose signals research

Researchers from MIT CSAIL have developed an AI capable of detecting movement through walls using just RF wireless signals. CSAIL (Computer Science and Artificial Intelligence Laboratory) is based at the Massachusetts Institute of Technology with the goal of ‘pioneering new approaches to computing that will bring about positive changes in the way people around the globe live, play, and work.’ The researchers’ latest development, RF-Pose, uses a neural network in combination with simple...

Don’t be a(nother) Joker, let AI create a unique Halloween costume

As Halloween approaches, we’ll be seeing many Joker and Harley Quinn costumes dragged back out the depths of closets. Pennywise the clown is another obvious choice for this year. If you want something a bit more unique for a costume idea, an AI could help. Research scientist Janelle Shane decided to build a neural network fed with data of 4,500 costume names she crowdsourced from the internet (which is always a safe bet*) Speaking to Business Insider, Shane claims the network’s...

Prowler.io aspires to build AI which makes human-like decisions

Cambridge-based AI startup Prowler has raised £10 million to help it build an AI which can make human-like decisions. Based on the comments made by Elon Musk and Vladimir Putin in our article yesterday, you’d be forgiven if this raises some concerns. AI like Prowler, however, could be what saves us from destruction. If you missed it, Musk voiced his concern about Putin’s comment that the nation which leads in AI “will become the ruler of the world.” Some are concerned about an...