That’s helpful because GPs are mathematically much simpler than NNs (but algorithmically more complex for large data).
@lawrennd
-
Gaussian Processes Explain Modern Neural Networks Generalization Behavior
By
–
GPs also give insights on why modern (very large) NNs behave like they do e.g the reason they generalise seems closer to non parametric approaches like GPs than traditional NN generalisation understanding from 1990s.
-
Chris’s Inspiration Expands GP Community Despite NN Growth
By
–
The point being that Chris inspired me (and many others). So he did his job. That means GP community today is larger than it’s ever been. It’s just that NN community grew even more. But both are GPs and NNs are successful.
-

ConvNet researcher’s luxury car versus Gaussian Process bicycle exit
By
–
The Convnet guy left the workshop in a Porsche whereas the GP guy left it on a bicycle. @bschoelkopf
-

Celebrating ELLIS Launch with Friends and Football
By
–
So nice to spend time with old and new friends celebrating the launch of
@ELLISInst_Tue
but also watching Germany play football with
@ylecun
Sepp Hochreiter
@bschoelkopf
. -
AI Deployment in Public Service Initiative Begins
By
–
It started as frontier task force for deploying AI in public service.
-
AISI’s role and multiple strategic pivots questioned
By
–
My understanding is/was that AISI is not a policy making body. Although it's pivoted more times than a folding ruler so that may have changed.
-
Open Source Ecosystem Support Through Policy Beyond Legislation
By
–
And yet they ensured their open source ecosystem is supported. Policy isn’t just legislation.
-
Existing AI Legislation Applies: UK Sector-Led Approach Defended
By
–
This is simply untrue. Much existing legislation applies to AI (thus the UK sector led approach). Although there are certainly gaps and overlaps (thus need for coordination).
-
UK AI Council Loss Enables Regulatory Capture, Stifles Open Source
By
–
The loss of the UK AI Council and other advisory structures has left a void of expertise in UK AI Policy. That has enabled regulatory capture, which has stymied development of open source AI ecosystem in the UK. Contrast with France is stark.