Models for low-latency, streaming applications could benefit from the knowledge capacity of larger models, but edge devices cannot run these models due to resource constraints. A possible solution is to transfer hints during inference from a large model running remotely to a small model running on-device. However, this incurs a communication delay that breaks real-time requirements and does not guarantee that both models will operate on the same data at the same time. We propose knowledge boosting, a novel technique that allows a large model to operate on time-delayed input during inference, while still boosting small model performance. Using a streaming neural network that processes 8 ms chunks, we evaluate different speech separation and enhancement tasks with communication delays of up to six chunks or 48 ms. Our results show larger gains where the performance gap between the small and large models is wide, demonstrating a promising method for large-small model collaboration for low-latency applications. Code, dataset, and audio samples available at https://knowledgeboosting.cs.washington.edu/.

Prototype noise-cancelling headphones allow you to select which background noises to drown out, letting you put an "audio spotlight" on one specific voice so you can concentrate on it. Conventional noise-cancelling headphones reduce unwanted sounds like the rumble of a bus engine, but because the technology cancels out certain frequencies entirely, it can also suppress sounds we want to hear. Now, Shyam Gollakota at the University of Washington in Seattle and his colleagues have created headphones that can remove any unwanted noises while leaving others intact, regardless of their frequencies. It can also be trained with the press of a button to home in on a specific person's voice and exclude all other noise. The researchers are presenting their prototype at a joint meeting of the Acoustical Society of America and the Canadian Acoustical Association this week.

When NVIDIA purchased mobile-chip designer Arm Holdings from SoftBank last year, NVIDIA CEO Jensen Huang made the bold prediction that in the years ahead, there will be trillions of artificial intelligence (AI)-enabled Internet of Things (IoT) devices. Regardless of whether that holds true, it is safe to say the growth of IoT devices is exploding. All those devices will require power sources, and the way Josiah Hester sees it, that's problematic for the environment and society. "When I see the'trillion' number, I see a trillion dead batteries, basically," says Hester, an assistant professor of computer engineering at Northwestern University. "There's piles of batteries in landfills in China and elsewhere sitting there unrecycled; or they're put in furnaces and melted down, which is not a carbon-neutral event."

Researchers at the University of Washington have figured out a way to use machine-learning algorithms to turn smart speakers into sensitive medical devices that can detect irregular heartbeats. The scientists use smart speakers like Amazon Echo or Google Home to send out an inaudible sound that bounces off a person's chest and returns to the device, reshaped in a way that reveals the heartbeat. An uneven cardiac rhythm can be associated with ailments including strokes or sleep apnea. The researchers employed a machine-learning algorithm to tease out the heartbeats from other sounds and signals such as breathing, which is easier to detect because it involves a much larger motion. The algorithm was also needed to zero in on erratic heart rhythms -- which from a health perspective are generally more important to identify than a steady "lub-dub."

Shyam Gollakota and his colleagues at the University of Washington in the US have developed a small steerable camera that can be affixed to beetles to transmit footage from their surrounding environment in real time. The camera uses Bluetooth to stream footage to a smartphone at a resolution of 160 by 120 pixels, and at a rate of between one and five frames per second. It sits on a mechanical arm that can be remotely controlled to pivot the camera frame left and right. Capturing footage while beetles move has a power-saving advantage over insect-like robots or drones, says Gollakota. "That mobility really drains the battery a lot," he says.

Washington: Scientists have developed a new artificial intelligence (AI) system to monitor people for cardiac arrest while they are asleep without touching them. People experiencing cardiac arrest will suddenly become unresponsive and either stop breathing or gasp for air, a sign known as agonal breathing, said rese-archers at the University of Washington (UW) in the US. A new skill for a smart speaker -- like Google Home and Amazon Alexa -- or smartphone lets the device detect the gasping sound of agonal breathing and call for help. Immediate Cardiop-ulmonary resuscitation (CPR) can double or triple someone's chance of survival, but that requires a bystander to be present. CPR is an emergency procedure that combines chest compressions often with artificial ventilation in an effort to manually preserve intact brain function. Recent research suggests that one of the most common locations for an out-of-hospital cardiac arrest is in a patient's bedroom, where no one is likely around or awake to respond and provide care.

A New Jersey woman is alive because her Apple Watch alerted her to an elevated heart rate. It turned out she had fluid around her heart from a viral infection. Medical alert systems have been around for some time. Often, they're wearable devices that can detect when you fall, and alert emergency personnel if it senses you aren't responding. But what happens if you aren't wearing a device, or if you aren't experiencing any triggering signs or symptoms of a medical emergency at all?

The research was led by Justin Chan, a PhD student in the department of computer science and engineering. Almost 500,000 Americans die each year from a cardiac arrest, the researchers wrote in the journal npj Digital Medicine. And the condition kills 100,000 Britons annually, according to Arrhythmia Alliance. Study author Dr Jacob Sunshine, assistant professor of anesthesiology and pain medicine, said: 'Cardiac arrests are a very common way for people to die and right now many of them can go unwitnessed. 'Part of what makes this technology so compelling is that it could help us catch more patients in time for them to be treated.'

Insect-sized flying robots could help with time-consuming tasks like surveying crop growth on large farms or sniffing out gas leaks. These robots soar by fluttering tiny wings because they are too small to use propellers, like those seen on their larger drone cousins. Small size is advantageous: These robots are cheap to make and can easily slip into tight places that are inaccessible to big drones. But current flying robo-insects are still tethered to the ground. The electronics they need to power and control their wings are too heavy for these miniature robots to carry.

But current flying robo-insects are still tethered to the ground. The electronics they need to power and control their wings are too heavy for these miniature robots to carry. Now, engineers at the University of Washington have for the first time cut the cord and added a brain, allowing their RoboFly to take its first independent flaps. This might be one small flap for a robot, but it's one giant leap for robot-kind. The team will present its findings May 23 at the International Conference on Robotics and Automation in Brisbane, Australia.