Day: October 2, 2019

Generative Classifier for Detecting Out-of-Distribution and for handling noisy lables

A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks Preliminary 우선 Supplementary A에 대한 간단한 리뷰가 필요해 보임. Classifier는 크게 discriminative 와 generative classifier 두 가지로 나뉨. discriminative는 x가 주어졌을 때 어떤 class(y)인지 직접적으로 예측함, 즉 P(y|x). 대표적인게 softmax classifier이고 posterior distribution이 다음과 같이 정의됨. genertive classifier는 posterior distribution P(y|x) = P(x|y)P(y)/P(x)를 구하기 […]

Read more

Few-Shot Adversarial Learning of Realistic Neural Talking Head Models

Preprocess K+1 frames K : train Embedder [B(batch) * K, 2, C, W, H] x : frame [B * K, C, W, H] y : landmark [B * K, C, W, H] 1(t frame) : train Generator e_hat = E(x, y) x_hat = G(y_t, e_hat) r_x_hat = D(x_hat, y_t, i) r_x = D(x_t, y_t, i) LossEG […]

Read more