Timezone: »

Learning large-margin halfspaces with more malicious noise
Phil Long · Rocco A Servedio

Wed Dec 14 08:45 AM -- 02:59 PM (PST) @

We describe a simple algorithm that runs in time poly(n,1/gamma,1/eps) and learns an unknown n-dimensional gamma-margin halfspace to accuracy 1-eps in the presence of malicious noise, when the noise rate is allowed to be as high as Theta(eps gamma sqrt(log(1/gamma))). Previous efficient algorithms could only learn to accuracy eps in the presence of malicious noise of rate at most Theta(eps gamma). Our algorithm does not work by optimizing a convex loss function. We show that no algorithm for learning gamma-margin halfspaces that minimizes a convex proxy for misclassification error can tolerate malicious noise at a rate greater than Theta(eps gamma); this may partially explain why previous algorithms could not achieve the higher noise tolerance of our new algorithm.

Author Information

Phil Long (NEC Labs)
Rocco A Servedio (Columbia University)

More from the Same Authors