The trilemma of neural nets.
ACI holds all three.
ACI is the result of years of research into neural nets and their trilemma: plasticity, stability and editability.
Two at best, for every neural net.
Learn faster and it forgets. Hold it steady and it stops learning. And what it learned is spread through all of it, so no one piece can be changed. Every gain on one costs another.
Step 1 of 5: Plasticity
Plasticity: learn the new, fast.
A network that learns fast overwrites what it knew. Researchers call it catastrophic forgetting. ACI learns from a single confirmed result, and the very next request has it.
Step 1 of 4: Something new
Your team confirms something new. A network needs a training run to take it in. ACI takes it in from that one result.
Stability: keep what it knows.
Hold a network steady and it keeps what it knows, but it stops learning. ACI goes on learning, and everything it learned before stays exactly as it was, for as long as it runs.
Step 1 of 4: Hold it steady
Freeze a network to keep what it knows, and it stops learning. ACI goes on learning, and everything it knew stays as it was.
Editability: change what it learned.
What a network learned is spread through all of it, so one thing can’t be corrected or removed without retraining. Anything ACI learned can be reviewed, corrected or removed on its own, and the rest stays as it was.
Step 1 of 4: Find it
Ask where a network learned something, and no one can point to it. With ACI, you can see where every piece came from.
All three, at runtime.
Holding all three is what lets ACI learn and act at runtime, from your work, for as long as it runs. It is what makes ACI a new category of intelligence.
With all three, an intelligence can learn from every result while the work happens, build on everything it learned before, and stay in your hands: anything it learned can be reviewed, corrected or removed.
That is what continual learning has always asked for, and what a neural net on its own can’t give. ACI gives it to the AI models you already use, in tandem with them.
Years of research, a new kind of intelligence.
ACI is the result of years of research into neural nets and their trilemma: plasticity, stability and editability.
It started with three questions every neural net raises: why it forgets when it learns, why it stops learning once it is steady, and why nothing it has learned can be changed on its own.
The answers became ACI: Vareon’s proprietary algorithms and architecture for an intelligence that learns from every result, builds on what it knows and stays in your hands. Duplex brings it to your everyday work.
See Duplex on your own work.
Duplex is built on ACI. A pilot puts it on your own work, with the measures agreed before it begins.