Timezone: »

Efficient Second Order Online Learning by Sketching
Haipeng Luo · Alekh Agarwal · Nicolò Cesa-Bianchi · John Langford

Tue Dec 06 09:00 AM -- 12:30 PM (PST) @ Area 5+6+7+8 #17

We propose Sketched Online Newton (SON), an online second order learning algorithm that enjoys substantially improved regret guarantees for ill-conditioned data. SON is an enhanced version of the Online Newton Step, which, via sketching techniques enjoys a running time linear in the dimension and sketch size. We further develop sparse forms of the sketching methods (such as Oja's rule), making the computation linear in the sparsity of features. Together, the algorithm eliminates all computational obstacles in previous second order online learning approaches.

Author Information

Haipeng Luo (Princeton University)
Alekh Agarwal (Microsoft)
Nicolò Cesa-Bianchi (Università degli Studi di Milano, Italy)
John Langford (Microsoft Research New York)

More from the Same Authors