3. 复习

3.1. 汇总

1. github

• https://github.com/imhuay/Algorithm_Interview_Notes-Chinese

• https://github.com/jwasham/coding-interview-university/blob/master/translations/README-cn.md

2. 2018校招算法岗面试题汇总

https://zhuanlan.zhihu.com/p/36801851

3.2. C++

1. 虚函数

https://blog.csdn.net/fighting_coder/article/details/77187151

2. C++构造函数和析构函数能否声明为虚函数？(转载)

https://www.cnblogs.com/hxb316/p/3853544.html

3. 重载、重写（覆盖）和隐藏的区别

https://blog.csdn.net/zx3517288/article/details/48976097

4. C++ STL中vector内存用尽后，为啥每次是两倍的增长，而不是3倍或其他数值？Hint：\(1 + 2 + 1 + 4 + 1 + 1 + 1 + 8 + \cdots + n = \mathcal{O}(n)\) ，每一次 push_back 操作的摊还代价为 \(\mathcal{O}(1)\) 。

https://www.zhihu.com/question/36538542

5. 常见C++笔试面试题整理

https://zhuanlan.zhihu.com/p/69999591

3.3. Python

1. 基本数据类型

https://www.cnblogs.com/littlefivebolg/p/8982889.html

2. Python中的None

https://www.cnblogs.com/changbaishan/p/8084863.html

3. 使用lambda高效操作列表的教程

https://www.cnblogs.com/mxp-neu/articles/5316557.html

4. 经典7大Python面试题

https://blog.csdn.net/qq_41597912/article/details/81459804

5. 迭代器和生成器

https://www.cnblogs.com/chongdongxiaoyu/p/9054847.html

3.4. 机器学习/深度学习

1. 激活函数

https://fongyq.github.io/blog/deepLearning/02_activationFunction.html

2. Batch Normalization

https://fongyq.github.io/blog/deepLearning/03_batchnorm.html

3. 过拟合

https://fongyq.github.io/blog/deepLearning/03_batchnorm.html

4. 正则化项L1和L2的区别

https://www.cnblogs.com/lyr2015/p/8718104.html

5. KMeans秘籍之如何确定K值

https://blog.csdn.net/alicelmx/article/details/80991870

6. 决策树

• ID3、C4.5

  https://www.cnblogs.com/coder2012/p/4508602.html

• 预剪枝与后剪枝

  https://blog.csdn.net/zfan520/article/details/82454814

• CART分类与回归树

  https://www.jianshu.com/p/b90a9ce05b28

7. Logistic Regression

https://fongyq.github.io/blog/machineLearning/01_lr.html

8. Support Vector Machine

https://fongyq.github.io/blog/machineLearning/02_svm.html

9. PCA

https://fongyq.github.io/blog/machineLearning/03_pca.html

3.5. 论文相关

1. AlexNet/VGG/GoogleNet

https://blog.csdn.net/gdymind/article/details/83042729

2. CNN卷积神经网络_ GoogLeNet 之 Inception(V1-V4)

https://blog.csdn.net/diamonjoy_zone/article/details/70576775

3. ResNet

• Deep Residual Learning for Image Recognition

  https://arxiv.org/pdf/1512.03385.pdf

• torchvision.models.resnet

  https://pytorch.org/docs/stable/_modules/torchvision/models/resnet.html#resnet101

• ResNet-50 结构

  https://www.jianshu.com/p/993c03c22d52

4. ResNeXt

• ResNeXt

  https://www.cnblogs.com/bonelee/p/9031639.html

• ResNeXt算法详解

  https://blog.csdn.net/u014380165/article/details/71667916

5. R-CNN系列

• RCNN（三）：Fast R-CNN

  https://blog.csdn.net/u011587569/article/details/52151871

• 一文读懂Faster RCNN

  https://zhuanlan.zhihu.com/p/31426458

• 【RCNN系列】【超详细解析】

  https://blog.csdn.net/amor_tila/article/details/78809791

• 实例分割模型Mask R-CNN详解：从R-CNN，Fast R-CNN，Faster R-CNN再到Mask R-CNN

  https://blog.csdn.net/jiongnima/article/details/79094159

• (Mask RCNN)——论文详解(真的很详细)

  https://blog.csdn.net/wangdongwei0/article/details/83110305

• 实例分割–Mask RCNN详解（ROI Align / Loss Fun）

  https://blog.csdn.net/qinghuaci666/article/details/80900882

• ROI-Align 原理理解

  https://blog.csdn.net/gusui7202/article/details/84799535

• 为什么RCNN用SVM做分类而不直接用CNN全连接之后softmax输出？

  https://www.zhihu.com/question/54117650

6. Focal Loss（样本不均衡：正/负样本数量不均衡（ \(\alpha\) ），简单/困难样本数量不均衡（ \(\gamma\) ））

\[\begin{split}CE &=\ -y \log y_t - (1 - y) \log (1 - y_t) & &\ [\text{Cross Entropy Loss}] \\ FL &=\ -y \alpha (1 - y_t)^\gamma \log y_t - (1 - y) (1 - \alpha) y_t^\gamma \log (1 - y_t) & &\ [\text{Focal Loss}]\end{split}\]

即

$$ CE = \begin{cases} - \log y_t, & &\ y=1\\ - \log (1 - y_t), & &\ y=0 \end{cases} $$ $$ FL = \begin{cases} - \alpha (1 - y_t)^\gamma \log y_t, & &\ y=1\\ - (1 - \alpha) y_t^\gamma \log (1 - y_t), & &\ y=0 \end{cases} $$

• 损失函数改进方法之Focal Loss

  https://blog.csdn.net/sinat_24143931/article/details/79033538

• RetinaNet论文理解

  https://blog.csdn.net/wwwhp/article/details/83317738

• Focal Loss理解

  https://www.cnblogs.com/king-lps/p/9497836.html

7. FCN（Fully Convolutional Networks）

• FCN学习:Semantic Segmentation

  https://zhuanlan.zhihu.com/p/22976342?utm_source=tuicool&utm_medium=referral

• FCN于反卷积(Deconvolution)、上采样(UpSampling)

  https://blog.csdn.net/nijiayan123/article/details/79416764

8. FPN（Feature Pyramid Networks for Object Detection）

https://www.cnblogs.com/fangpengchengbupter/p/7681683.html

9. CapsuleNet解读

https://blog.csdn.net/u013010889/article/details/78722140/

10. 轻量级网络–MobileNet论文解读

https://blog.csdn.net/u011974639/article/details/79199306

11. 一文读懂卷积神经网络中的1x1卷积核。Hint：升维、降维、跨通道信息交互。

https://zhuanlan.zhihu.com/p/40050371

12. Image Classification Architectures

• 模型，FLOP（浮点计算量），性能，参数量（表格中的参数量单位是字节，按 4 字节/浮点型数计算，需要除以 4 才得到参数个数）

  https://github.com/albanie/convnet-burden#convnet-burden

• torchvision.models

  https://pytorch.org/docs/stable/torchvision/models.html

3.6. 其他

1. 理解数据库的事务，ACID，CAP和一致性

https://www.jianshu.com/p/2c30d1fe5c4e