AI Challenger 2018 细粒度用户评论情感分类第一名解决方案/阅读理解解决方案 https://github.com/chenghuige/wenzheng https://mp.weixin.qq.com/s/W0PhbE8149nD3Venmy33tw
Kaggle PLAsTiCC天文分类比赛第四名方案 https://github.com/aerdem4/kaggle-plasticc https://www.kaggle.com/c/PLAsTiCC-2018/discussion/75011
有关语义分割的奇技淫巧有哪些 https://www.zhihu.com/question/272988870 https://github.com/liaopeiyuan/ml-arsenal-public
参加Kaggle涂鸦识别比赛的十条经验教训 https://towardsdatascience.com/10-lessons-learned-from-participating-to-google-ai-challenge-268b4aa87efa
技术面试常见问题/解答集锦 https://github.com/FAQGURU/FAQGURU
'船长黑板报 - 关于机器学习、计算机视觉和工程技术的总结和分享' https://github.com/Captain1986/CaptainBlackboard
GBDT 和 Random Forest 为什么在Kaggle等比赛中效果非常好 https://bigquant.com/community/t/topic/259?suanfazu&181206&L3
机器学习项目失败的9个原因 https://mp.weixin.qq.com/s/SWuaM7PqOD4MK_rpnGaB5w
在文本分类任务中，有哪些论文中很少提及却对性能有重要影响的tricks？ https://www.zhihu.com/question/265357659
技术面试参考资料大列表 https://github.com/Awesome-Interview/Awesome-Interview
Bringing Giant Neural Networks Down to Earth with Unlabeled Data https://arxiv.org/abs/1907.06065
Towards Explaining the Regularization Effect of Initial Large Learning Rate in Training Neural Networks https://arxiv.org/abs/1907.04595
为什么机器学习模型在产品化过程中会退化？ https://medium.com/m/global-identity?redirectUrl=https%3A%2F%2Ftowardsdatascience.com%2Fwhy-machine-learning-models-degrade-in-production-d0f2108e9214
【如何扩展深度网络的能力】《How to trick deep learning algorithms into doing new things | TechTalks》 https://bdtechtalks.com/2020/07/20/black-box-adversarial-reprogramming/
Black Magic in Deep Learning: How Human Skill Impacts Network Training https://arxiv.org/abs/2008.05981
数据科学失败案例集锦 https://github.com/xLaszlo/datascience-fails
算法工程师必备的几种调参技艺 https://medium.com/swlh/4-hyper-parameter-tuning-techniques-924cb188d199
Source-Code-Notebook - 关于一些经典论文源码的逐行中文笔记(推荐系统/图神经网络/计算机视觉/自然语言处理/跨模态)’ https://github.com/nakaizura/Source-Code-Notebook
《Bag of Tricks for Long-Tailed Visual Recognition with Deep Convolutional Neural Networks》(AAAI 2021) github.com/zhangyongshun/BagofTricks-LT
如何让炼丹更有条理 github.com/ahangchen/windy-afternoon/blob/master/ml/pratice/torch_best_practice.md