The discovery that led Nir Shavit to start a company came about the way most discoveries do: by accident. The MIT professor was working on a project to reconstruct a map of a mouse’s brain and needed some help from deep learning. Not knowing how to program graphics cards, or GPUs, the most common hardware choice for deep-learning models, he opted instead for a central processing unit, or CPU, the most generic computer chip found in any average laptop.

“Lo and behold,” Shavit recalls, “I realized that a CPU can do what a GPU does—if programmed in the right way.”

This insight is now the basis for his startup, Neural Magic, which launched its first suite of products today. The idea is to allow any company to deploy a deep-learning model without the need for specialized hardware. It would not only lower the costs of deep learning but also make AI more widely accessible.

The search giant was the only Chinese member of the Partnership on Artificial Intelligence, a US-led effort to foster collaboration on ethical issues.

Artificial intelligence networks have learnt a new trick: being able to create photo-realistic faces from just a few pixelated dots, adding in features such as eyelashes and wrinkles that can’t even be found in the original.

Before you freak out, it’s good to note this is not some kind of creepy reverse pixelation that can undo blurring, because the faces the AI comes up with are artificial – they don’t belong to real people. But it’s a cool technological step forward from what such networks have been able to do before.

The PULSE (Photo Upsampling via Latent Space Exploration) system can produce photos with up to 64 times greater resolution than the source images, which is 8 times more detailed than earlier methods.

In recent years, many research teams worldwide have been developing and evaluating techniques to enable different locomotion styles in legged robots. One way of training robots to walk like humans or animals is by having them analyze and emulate real-world demonstrations. This approach is known as imitation learning.

Researchers at the University of Edinburgh in Scotland have recently devised a framework for training humanoid robots to walk like humans using human demonstrations. This new framework, presented in a paper pre-published on arXiv, combines imitation learning and deep reinforcement learning techniques with theories of robotic control, in order to achieve natural and dynamic locomotion in humanoid robots.

“The key question we set out to investigate was how to incorporate useful human knowledge in robot locomotion and human motion capture data for imitation into deep reinforcement learning paradigm to advance the autonomous capabilities of legged robots more efficiently,” Chuanyu Yang, one of the researchers who carried out the study, told TechXplore. We proposed two methods of introducing human prior knowledge into a DRL framework.”