c57ce7f891a91592354b1685d37f2aee39f193f2,libact/query_strategies/active_learning_by_learning.py,Exp4P,exp4p,#Exp4P#,160

Before Change

            ask_idx = self.invert_id_idx[ask_id]

            rhat[ask_idx] = reward / p[ask_idx]
            for i in range(self.N):
                yhat[i] = np.dot(advice[i], rhat)
                vhat[i] = np.sum(advice[i] / p)

                self.w[i] = self.w[i] * np.exp(
                        self.pmin / 2 * (yhat[i] + vhat[i]*np.sqrt(
                            np.log(self.N/self.delta) / self.K / self.T)
                            )
                        )

            self.t += 1

        raise StopIteration

After Change

            //                np.log(self.N/self.delta) / self.K / self.T)
            //                )
            //            )
            yhat = np.dot(advice, rhat)
            vhat = np.sum(advice / np.tile(p, (self.N, 1)), axis=1)
            self.w = self.w * np.exp(
                    self.pmin / 2 * (yhat + vhat*np.sqrt(
                        np.log(self.N/self.delta) / self.K / self.T)

Instances