News Quantum machine learning unlocks new efficient chip design pipeline — encoding data in quantum states then analyzing it with machine learning up to...

[Admin]

Researchers have discovered a new method of designing semi-conductors by modeling the electrical resistance inside a chip using quantum computing pattern recognition, and then machine learning to analyze the output data.

Quantum machine learning unlocks new efficient chip design pipeline — encoding data in quantum states then analyzing it with machine learning up to... : Read more

 

[Mr Majestyk]

"Although we might not be quite ready to revolutionize chip making with the techniques outlined in this study"

We are ready to hype the crap out this before we even have actual results.

 

[AIateMyHomework]

Quantum Kernel-Aligned Regressor (QKAR), which converts classic data into quantum states. The quantum computing system could then analyze that data to look for patterns in it.... As a human who just unlocked opposable thumbs (still pretty glitchy), I’m struggling to type this without smashing 3 keys at once. But wow… this Quantum Kernel-Aligned Regressor thing sounds amazing. If only I could harness its power to stop autocorrect from turning my words into gibberish every time my new thumbs go rogue…

 

[bit_user]

"Although we might not be quite ready to revolutionize chip making with the techniques outlined in this study"

We are ready to hype the crap out this before we even have actual results.

I doubt we'll ever learn about production uses of this technology. That would be proprietary and something semiconductor fabs would guard very closely (i.e. trade secret).

The study does compare with the solutions produced using other analytic methods:

"The study suggests QKAR was a more effective method than other models by between 8.8% and 20.1%."
​

They picked just one optimization problem that's prevalent in semiconductor manufacturing, but I'm sure there are somewhere between dozens and hundreds involved in the design of each process node. This is probably one of the reasons Intel was churning through so many wafers in the development of the High-NA node:

https://www.tomshardware.com/tech-i...0-000-wafers-with-high-na-euv-chipmaking-tool
​

 

[bit_user]

Quantum Kernel-Aligned Regressor (QKAR), which converts classic data into quantum states. The quantum computing system could then analyze that data to look for patterns in it....

If I understand correctly, the key development is that you don't need to develop a mathematical model for the quantum system to solve. It's a more direct way of just plugging the data and letting the quantum computer find a model and then optimize it.

 

[AIateMyHomework]

Yes, correct , however for larger models it becomes unstable , so we need a way of sharing these models.