A team of researchers from the University of Oxford, in collaboration with other universities, has developed an innovative hardware system that combines photonic and electronic technologies to process 3D data. The system significantly enhances processing power for AI tasks. To test the hardware, the team analyzed 100 electrocardiogram signals simultaneously and achieved a 93.5% accuracy rate in identifying the risk of sudden death. The researchers believe that this approach could lead to a 100-fold increase in energy efficiency and compute density compared to current electronic processors if scaled up.

The efficiency of conventional computer chip processing doubles every 18 months. However, modern AI tasks require processing power that is currently doubling every 3.5 months. This means that new supercomputing paradigms are urgently needed to cope with this rising demand.

One possible solution is to use light instead of electronics to carry out multiple calculations in parallel using different wavelengths to represent different sets of data. In 2021, the same authors published groundbreaking work demonstrating a form of integrated photonic processing chip that could carry out matrix-vector multiplication at a much faster speed than the fastest electronic approaches. This breakthrough led to the creation of Salience Labs, a photonic AI company that emerged from the University of Oxford.

The team has now taken this concept further by adding an extra parallel dimension to the processing capability of their photonic matrix-vector multiplier chips. This higher-dimensional processing is made possible by using multiple different radio frequencies to encode the data, thereby achieving a level of parallelism that was previously impossible.

The team tested the hardware by applying it to the task of assessing the risk of sudden death from electrocardiograms of heart disease patients. They were able to successfully analyze 100 electrocardiogram signals simultaneously, accurately identifying the risk of sudden death with 93.5% accuracy.

The researchers estimated that even with a moderate scaling of 6 inputs x 6 outputs, this approach could outperform state-of-the-art electronic processors, potentially providing a 100-times enhancement in energy efficiency and compute density. The team anticipates further enhancement in supercomputing parallelism in the future by exploiting more degrees of freedom of light, such as polarization and mode multiplexing.

Dr. Bowei Dong, the first author of the publication, expressed his gratitude for the vibrant and collaborative platform provided by Oxford, which gave him the opportunity and courage to push the frontiers of advanced AI supercomputing hardware. Professor Harish Bhaskaran, the co-founder of Salience Labs and leader of this work, said that this is an exciting time to be doing research in AI hardware at the fundamental scale, and this work is one example of how what we assumed was a limit can be further surpassed.

The Wild Salmon Center has partnered with several First Nations to use a combination of cutting-edge artificial intelligence tools and traditional Indigenous fishing methods to gain a better understanding of salmon runs in real time. The Salmon Vision deep learning model, which uses advanced artificial intelligence tools to identify and count fish species, is currently being utilized in various rivers around the North and Central Coasts of British Columbia. By 2024, Salmon Vision aims to provide reliable real-time fish count data to First Nations fisheries managers, thereby increasing their involvement in fisheries management decisions.

Fisheries managers on British Columbia’s Central Coast have to make decisions without knowing how many salmon are returning until after fishing seasons are over. They have to make forecasts and set harvest targets for commercial and recreational fisheries based on modeled data from the past. Emergency closures also have to be decided on when salmon populations start to decline. However, with the unpredictable and accelerating effects of climate change, it is increasingly difficult to rely on past data to predict future salmon returns. 

Dr. Will Atlas, Wild Salmon Center Senior Watershed Scientist, suggests a solution called “Salmon Vision.” A first-of-its-kind technology that combines artificial intelligence with ancient fishing weir technology, the Salmon Vision computer deep learning model can identify and count fish species. Developed by WSC in data partnership with the Gitanyow Fisheries Authority and Skeena Fisheries Commission, Salmon Vision aims to enable real-time salmon population monitoring for First Nations fisheries managers and beyond.

Automating fish counting is crucial for informed decisions while salmon are still running, according to many of our First Nations partners. Dr. Atlas suggests that underwater video technology can help us see those salmon returning to rivers. 

The Salmon Vision pilot study has annotated over 500,000 video frames captured at Indigenous-run fish counting weirs on the Kitwanga and Bear Rivers of B.C.'s Central Coast. Early assessments indicate that the technology is adept at tracking 12 different fish species passing through custom fish-counting boxes at the two weirs, with scores surpassing 90 and 80 percent accuracy for coho and sockeye salmon: two of the principal fish species targeted by First Nations, commercial, and recreational fishers. 

The Heiltsuk Nation is running Salmon Vision on a weir on the Koeye River. For First Nations like the Heiltsuk, weirs represent more than a revitalization of an age-old fishing technology. The rebuilding of weirs on rivers like the Koeye is a statement of First Nations sovereignty and their seat at the table in fisheries management decisions, as they were banned in the late 1800s by Canada's Department of Fisheries and Oceans as a way to consolidate control of fishery resources. 

"Modern-day expression of Heiltsuk title and rights and an avenue for us to be a part of the latest science," says William Housty, Associate Director of the Heiltsuk Integrated Resource Management Department. "And to make decisions not just for the betterment of this creek, but for the whole ecosystem." 

The Salmon Vision team is implementing automated counting on a trial basis in several rivers around the B.C. North and Central Coasts with partner First Nations. The goal is to provide reliable real-time fish count data to these partners by 2024. Ultimately, Dr. Atlas says, this groundbreaking A.I. technology could be in place in rivers across the North Pacific. 

"How many salmon are returning everywhere that we're fishing for salmon is the information we need," Dr. Atlas says. "You can't tell me with a straight face that you're having a sustainable fishery if you don't know how many fish you have coming back. And that's a problem right around the Pacific Rim." 

It's a problem with a promising solution, one that's just now coming into focus.