論文閱讀記錄

04-28

HEAD

2018-01-23

Global overview of Imitation Learning：imitation learning

2018-01-22

Piggyback: Adding Multiple Tasks to a Single, Fixed Network by Learning to Mask：continual learning

2017-12-20

Bessel Function：波的傳播，可以用來設計濾波器

Exponential family：一系列概率分布

Function Approximation

PassGAN: A Deep Learning Approach for Password Guessing

NAG: Network for Adversary Generation

mixup: Beyond Empirical Risk Minimization

矩陣分解思想應用可真廣泛，從降維（PCA）、信息檢索（LSI）、自然語言處理（LSA）到推薦系統。說到底其實就是通過線性變換將原始空間轉換到一個新空間。

生成模型擬合數據的probability density function，判別模型擬合數據的decision boundary。

2017-12-19

Hough Transform：知道形狀方程，尋找形狀

RANSAC：知道表達式，但是數據中雜訊很嚴重，確定表達式中的參數

防止擬合方法：label smooth、soft target、one sided label smooth

2017-12-18

MentorNet: Regularizing Very Deep Neural Networks on Corrupted Labels

Mathematics of Deep Learning

Relation Networks for Object Detection: attention used in cv

2017-12-14

count vector -> tf-idf -> lsa -> plsa -> lda

co-occurrence matrix -> svd分解 -> word2vec

LSA:PCA分解tf-idf矩陣

FM、NFM、PFM

LSA是代數思路解決問題，pLSA是概率思路解決問題，LDA是圖模型解決問題。

自然語言理解有兩個困難的問題，一詞多義，多詞同義

Deep Learning: Practice and Trends

Gaussian Processes: nonparametric

Dimensionality Reduction and Latent Topic Models

Unsupervised Learning – Topic Models

Introduction to Topic Models

Dimensionality Reduction and Topic Modeling

Comparing Latent Dirichlet Allocation and Latent Semantic Analysis as Classifiers

Topic Modeling

Latent Dirichlet Allocation vs Latent Semantic Indexing

LSA、pLSA、LDA

PLSI

Topic Model

LSA tutorial

Word2vec vs LDA

A Brief History of Word Embeddings

2017-12-13

卷積神經網路進展

Xception: Deep Learning with Depthwise Separable Convolutions

2017-12-12

An Intuitive Understanding of Word Embeddings: From Count Vectors to Word2Vec

Word Vectors (cs224d)

Distributed Representations (csc321)

Word2vec

Deep learning for nlp advancements and trends in 2017

2017-12-11

mixup: Beyond Empirical Risk Minimizatio

2017-12-07

Adversarial

Adversarial example research

Cleverhans

Intriguing properties of neural networks

Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Images

Explaining and Harnessing Adversarial Examples

One pixel attack for fooling deep neural networks

GAN

NIPS 2016 Tutorial: Generative Adversarial Networks

2017-12-05

Unsupervised Learning

Unsupervised Learning by Predicting Noise

Learning Feature Representations with K-means

A Text Detection System for Natural Scenes with Convolutional Feature Learning and Cascaded Classification

Convolutional Clustering for Unsupervised Learning

2017-12-04

L2 Regularization versus Batch and Weight Normalization

ICLR 2018 Open Review Papers

Adversarial Attack and Defense

Security and Privacy in Machine Learning

Adversarial Example Research

NIPS2017 Non Targeted Adversarial Attack

2017-12-01

Deep Image Prior (inpainting, restoration, super resolution, denoising)

Plug & Play Generative Networks: Conditional Iterative Generation of Images in Latent Space

Deep Learning and the Game of Go

Learning to Segment Every Thing

2017-11-29

Distilling a Neural Network Into a Soft Decision Tree

StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation