Date: | Wed, 3 Jun 2015 19:55:37 -0700 |
From: | Bret Victor |
Subject: | Re: Tokens for Shawn |
On Jun 3, 2015, at 6:28 PM, Robert M Ochshorn wrote:Current status of the token-recognizer (works better than it deserves to[0]):(link)-rmo[0] getting extremely lazy with feature extraction—this is doing SVM with 30K raw pixels (absolutely no color correction, white balance, &c).