Mignon from the UK has a brand new proposal for AI on the sidelines

This story is syndicated from the premium edition of PreSeed Now, a newsletter dedicated to the product, market and founding history of UK founded startups, so you can understand how they fit into what is happening in the world and the startup -ecosystem fit.

The renewed excitement around the potential of AI heading into 2023 brings with it concerns about how best to process all of the data required for operations. This is far from a new challenge, however, and next-generation AI chips are being developed in labs around the world to tackle the challenge in different ways.

One of the first startups we ever featured at PreSeed Now follows a “neuromorphic” approach, influenced by the human brain. From a different direction comes a brand new spinout from Newcastle University called mignon (so new it doesn’t have a website yet).

Discover the future of technology!

Visit us at the TNW conference on June 15th and 16th in Amsterdam

According to the CEO, Mignon has developed an artificial intelligence chipset Xavier Parkhouse-Parkerhas “on the order of 10,000x performance improvements over alternative neural network-based chips for classification tasks”

classification is essentially the process of figuring out what the AI ​​sees, hears, reads, etc. – the first step in understanding the world around it, whatever use case it is deployed for. Mignon’s chipset was designed to be used in edge computing as a “classification coprocessor” on devices rather than in the cloud.

Additionally, Parkhouse-Parker says Mignon’s chipset can also train AI models at the edge, meaning the models can be optimized for the specific, individual environments in which they are used.

A prototype design of Mignon’s Gen 1 chipset

A propositional suggestion

What gives Mignon its technology an edge over the competition is based on a less resource intensive approach propositional logic.

“Neural networks, the dominant algorithms in AI and machine learning today, typically require the execution of many layers of increasingly resource-intensive computations. They can take a long time to train and use, and can cost a lot of energy, and they also exist as a black box; You can’t explain why the algorithms came to a certain result,” says Parkhouse-Parker.

“Mignon is based on an algorithm that can be executed in a single layer, using propositional logic that maintains accuracy, but allows calculations to be performed much faster and consuming much less energy.”

And when it comes to launch, Mignon could also have a strong advantage.

“The investment and commercial scale required to succeed in the semiconductor industry is significant. One of the biggest challenges for many other competitors in this sector is that they rely on non-standard or “exotic” features that don’t easily scale within the current semiconductor manufacturing ecosystem,” said Parkhouse-Parker.

Instead, Mignon’s chipset uses a standard CMOS Manufacturing approach, meaning mass production is much easier.

How can it be used?

Edge AI has already made a tangible difference to consumers’ lives. Just look at how companies like Apple and Google have built AI chips into their smartphones to locally perform tasks like facial and object recognition in photos or audio transcription, increasing privacy and speed, and reducing data transfer costs.

According to Parkhouse-Parker, Mignon could ultimately make a difference here, along with the next generation of 6G telecom networks, where signal processing could be optimized by AI

But the first market they are targeting are industrial sectors where connectivity and power resources are low but powerful AI classification is required.

And while the technology isn’t ready for it, Parkhouse-Parker says Mignon is working on another selling point that will enable its offering – “explainable AI.” That means transparency about how and why AI made a certain decision.

To give a timely example if you ask OpenAI’s ChatGPT To explain a concept to you, you can’t understand why it comes to the specific answer that it is. You’ll only get a response based on the path it took through its sea of ​​data in response to your prompt.

In an industrial setting where AI may be making business-critical decisions or decisions that impact security, it would be very useful to be able to look back and see how the AI ​​came to the conclusion that it did it.

“With neural networks, all the reasoning is done inside a black box and you can’t see how or why this node is connected to this node or how things were calculated. Because Mignon is based on propositional logic, a researcher can see exactly where a decision was made, why, and what led them to that point,” explains Parkhouse-Parker.

Mignon wants to enable such accountability via software, which could be attractive in areas such as medicine, defense or the automotive industry.

The minds behind the Mignon product. LR: Professor Alex Yakovlev and Dr. Rishad Shafik

build mignon

Mignon technology comes from the work of Professor Alex Yakovlev And dr Rishad Shafik at Newcastle University.

Your research on taking the Tsetlin machine and incorporating it into computer hardware caught the attention of deep tech venture builders Cambridge Future Techwho also works with GitLife Biotech And imitationwhich have already been presented in this newsletter.

Since spring last year, Parkhouse-Parker (COO of Cambridge Future Tech) has been working on developing a commercial offer for Yakovlev and Shafik’s research. He has assumed the CEO role at Mignon as it is being spun off from the university.

Come on the market

First on the to-do list for the new startup is to further refine its technology with the development of a “Generation 2” chipset before bringing it to market.

“Although we’ve had fantastic performance improvements, and that’s actually quite remarkable, this was all done on the 65 nanometer node, which is old technology and should mean poorer performance improvements because the transistors are effectively larger, and that’s what we do.” truly remarkable,” says Parkhouse-Parker.

“We believe that if we move to a 28-nanometer node, all the numbers we have for the benchmarks will be significantly larger at that scale.”

Commercial validation is of course another important step after that. The ultimate goal is to work with fabless chip companies to integrate AA technology into a commercially available system-on-a-chip. Mignon has a number of settings planned for the near future to help him get there.

Mignon CEO Xavier Parkhouse-Parker

Investment plans and future potential

Parkhouse-Parker expect the spin-out process to be completed by March this year, after which they will officially open a £2.55million funding round.

This will be used to grow the team to design, test and manufacture the next generation of chipsets and to receive commercial validation across a range of industries. Software that enables AI development on the chipset is also a key part of the roadmap.

Finally, Parkhouse-Parker wants Mignon’s combination of low power consumption and broad compatibility to open up whole new possibilities for AI

“What Mignon is doing is a possibility for a really whole new world of devices that people haven’t even thought of. Think of the possibilities that would exist with product people like Steve Jobs or Jony Ive taking advantage of this and unleashing the potential. I think there really is a whole new world of possibilities.”

The big “hump”

There is no clear path from where Mignon is now to that future. Aside from the additional development work to refine the chipset, a rethink is needed from AI application developers.

“The big ‘hump,’ as one of our consultants calls it, is that it’s a new breed of artificial intelligence,” says Parkhouse-Parker. “The transition between neural networks and Tsetlin isn’t incredibly significant, but it will require a small mindset shift. It may require new ways of thinking about how artificial intelligence problems can be designed and how these things can be brought to market.

“There’s already a great community being built around it, but that’s one of the biggest challenges – building a Tsetlin ecosystem and moving things that are neural networks into Tsetlin.”

But despite these challenges, Parkhouse-Parker believes Mignon’s vision is entirely achievable.

“Improvements of several orders of magnitude warrant a look at something new, novel and exciting.”

The article you have just read is from the Premium Edition of PreSeed Now. This is a newsletter covering the product, market and history of startups created in the UK. The goal is to help you understand how these companies fit into what’s happening in the world and the startup ecosystem.

Comments are closed.