liuzihua_PKU_intern

DAY One : Try some fancy

(1) VAE

(2) autoencoder to reduce noise

(3)Word2Vec

(4) Style Variance

DAY Two : learning knowledge on K-means and EM Algorithm

(1) DCGAN face generaterations

(2)Predict the stocks

(3)K-means

(4) object_detections(google's API)

DAY Three : Use the dlib(a C++ module) to finish following cv tasks:

(1) Face detection

(2) Face key points detection

(3) Face recon

(4) Object blocks tracking

Plus:read the papar above on image caption by HKU 2018 ICCCV

DAY Four : User the paper read on day 4 and other reference to code a image caption project

image caption using encoder-decoder(LSTM with attendtion based)

DAY Five : make it clear for LSTM & Seq2Seq and attention-based model in SMT and apply it on Tensorflow .

(1) Neural Network Machine Translation(a Seq2Seq work)

(2) Neural Network Machine Translation(a Seq2Seq work using attention based model)

DAY Six : Review the concept of WGAN, CGAN， use instances to code the WGAN and the CGAN， write the intern daily of first week.

(1) WGAN face creation

(2) CGAN Mnist

DAY Seven : Focus on ACGAN,pixel to pixel ,and CycleGAN

(1) ACGAN

(2) CycleGAN

(3) Pixel to Pixel