0 写在前面
体悟:
- 从事人工智能行业,学术界产论文,工业界出产品.
- 工业应用上应用技术,能将GitHub的算法"跑通",能训练,能调参,或复现论文算法即可完成产品转化的第一步;
- 工程人员若想深入理解算法并进一步优化,则需要研究代码复现的论文,研究解决问题的原理及思路,这也是整理论文的意义所在;
- 当然,研究论文与"跑通"代码先后顺序依据个人或公司需求,若需要快速演示成果,则先跑通代码,若不急于展示,则先研读论文,再复盘代码;
- 做好笔记,可以是博客,在线文档,云盘等;
- 时间是检验真理的标准,实践也是检验真理的标准;
1 计算机视觉理论篇
二位大神佳作:
1.0 VGGNet
2 计算机视觉应用篇
2.0 图像风格转换
- "原始"方法Paper传送门
A Neural Algorithm of Artistic Style - "快速"方法Paper传送门
Perceptual Losses for Real-Time Style Transfer and Super-Resolution - 人脸转换Nvidia
A Style-Based Generator Architecture for Generative Adversarial Networks
2.1 图像修复
- Paper传送门
Context Encoders: Feature Learning by Inpainting - Paper传送门
High-Resolution Image Inpainting using Multi-Scale Neural Patch Systhesis
2.2 人脸识别
2.2 图像检测
- YOLO
You Only Look Once:Unified, Real-Time Object Detection - MR-CNN
Object detection via a multi-region & semantic segmentation-aware CNN model - Cascade R-CNN
Cascade R-CNN: Delving into High Quality Object Detection
2.3 视频实时目标检测
- Faster R-CNN
Faster R-CNN: Towards Real-Time Object
Detection with Region Proposal Networks - R-FCN
R-FCN: Object Detection via Region-based Fully Convolutional Networks - SSD
SSD: Single Shot MultiBox Detector - Mobile Devices
Pelee: A Real-Time Object Detection System on Mobile Devices - T-CNN
Quantization Mimic: Towards Very Tiny CNN for Object Detection
3 自然语言处理
3.1 句子分类池化
- 句子建模
A Convolutional Neural Network for Modelling Sentences - 句子分类
A Sensitivity Analysis of (and Practitioners’ Guide to) Convolutional Neural Networks for Sentence Classification - 句子分类
Convolutional Neural Networks for Sentence Classification
3.2 翻译
- Transforer模型
Attention Is All You Need - GLUE
GLUE: A Multi-Task Benchmark and Analysis Platform for Natural LanguageUnderstanding
- seqence2sequence
Sequence to Sequence Learning with Neural Networks
3.3 音乐识别
An Industrial-Strength Audio Search Algorithm
4 总结
- 多看论文,多写代码;
- 有坑,慢慢填;
- 戒骄戒躁;
- 相信时间,相信实践;
[参考文献]
[1]https://blog.csdn.net/yhl_leo/article/details/56674833
[2]https://cloud.tencent.com/developer/article/1162172
[3]https://handong1587.github.io/deep_learning/2015/10/09/object-detection.html#yolo