今天妙手偶得了一个非常好的机器学习平台End-to-End Machine Learning Library,这个平台是Brandon Rohrer创建的,里面用视频的方式介绍了机器学习的原理和应用。他的理念是:
The best way to learn new concepts is to use them to build something. These courses are structured to build foundational knowledge (100 series), provide in-depth applied machine learning case studies (200 series), and embark on project-driven deep-dives (300 series). Come have a look around and try one out today.
翻译成中文,就是:
学习新概念的最好方法是用它们来建造一些东西。这些课程的结构是建立基础知识(100系列),提供深入的应用机器学习案例研究(200系列),并开展项目驱动的深潜(300系列)。今天来看看,试一试吧。
对于机器学习的路径,他给出了关于神经网络、应用机器学习、数据可视化、数据科学等几个方面的推荐课程序列,非常值得借鉴。
Neural networks
193. How neural networks work
194. Build a neural network framework
195. Advanced neural network methods
196. Neural network optimization
Applied machine learning
197. How to choose a model
198. How optimization works
199. How selected models and methods work
200. Decision trees with Python an Pandas
201. Time-series analysis
202. Nonlinear modeling and optimization
Visualization
203. Navigating Matplotlib
204. Neural network visualization
Introduction to data science
206. Foundational skills
207. Data science concepts
208. Navigating a data science career