Jingfeng Wu
@uuujf.bsky.social
Postdoc at Simons at UC Berkeley; alumnus of Johns Hopkins & Peking University; deep learning theory.
https://uuujf.github.io
https://uuujf.github.io
sharing a new paper w/ Peter Bartlett, @jasondeanlee.bsky.social @shamkakade.bsky.social, Bin Yu
ppl talking about implicit regularization, but how good is it? We show it's surprisingly effective: GD dominates ridge for linear regression, w/ more cool stuff on GD vs SGD
arxiv.org/abs/2509.17251
ppl talking about implicit regularization, but how good is it? We show it's surprisingly effective: GD dominates ridge for linear regression, w/ more cool stuff on GD vs SGD
arxiv.org/abs/2509.17251
September 26, 2025 at 3:49 AM
sharing a new paper w/ Peter Bartlett, @jasondeanlee.bsky.social @shamkakade.bsky.social, Bin Yu
ppl talking about implicit regularization, but how good is it? We show it's surprisingly effective: GD dominates ridge for linear regression, w/ more cool stuff on GD vs SGD
arxiv.org/abs/2509.17251
ppl talking about implicit regularization, but how good is it? We show it's surprisingly effective: GD dominates ridge for linear regression, w/ more cool stuff on GD vs SGD
arxiv.org/abs/2509.17251
Sharing two new papers on accelerating GD via large stepsizes!
Classical GD analysis assumes small stepsizes for stability. However, in practice, GD is often used with large stepsizes, which lead to instability.
See my slides for more details on this topic: uuujf.github.io/postdoc/wu20...
Classical GD analysis assumes small stepsizes for stability. However, in practice, GD is often used with large stepsizes, which lead to instability.
See my slides for more details on this topic: uuujf.github.io/postdoc/wu20...
June 4, 2025 at 6:55 PM
Sharing two new papers on accelerating GD via large stepsizes!
Classical GD analysis assumes small stepsizes for stability. However, in practice, GD is often used with large stepsizes, which lead to instability.
See my slides for more details on this topic: uuujf.github.io/postdoc/wu20...
Classical GD analysis assumes small stepsizes for stability. However, in practice, GD is often used with large stepsizes, which lead to instability.
See my slides for more details on this topic: uuujf.github.io/postdoc/wu20...
