基于MPICH的KNN并行化计算，固定5个进程同时计算，经24W个训练样本学习，预测效果良好.-MPICH-based KNN parallel computing, fixed five processes simultaneously calculated by 24W training samples to learn, good prediction.

Hadoop隐马尔科夫模型的实现，有训练算法（EM）,产生的模型参数可以用来单机使用，亲测可用，可供正在学习hadoop的同学参考-Implement Hadoop hidden Markov model, there are training algorithm (EM), the model parameters can be used to produce stand-alone use, pro-test available for reference students are lear

利用多GPU进行分布式训练,用的是tensorflow的Keras平台(Distributed Training Using Multi-GPU)