Brief definition
1.从数据中学习( Learn from data)
More general definition:
1.没有显性编程设计过地学习(Learn without explicitly programmed)
2.能够对于发布的任务T,在某种性能衡量标准P下,从经验E中能够不断学习( learn from experience E, with respect to task T, under the performance measure P)
it can be classified based on several criteria:
1.是否有监督(whether or not they are trained with human supervision)
- K近邻(k-Neares Neighbors)
- 线性回归(Linear Regression)
- 逻辑斯特回归(Logistic Regression) : (can be used for classifiction可用于分类)
- 支持向量机(Support Vector Machines)
- 决策树和随机森林(Decison Tree and Random Forests)
- 神经网络(Neural Networks(some can be unsupervised))
- Clustering : 聚类
- - k-Means(k均值)
- - Hierarchical Cluster Analysis(HCA)(层次聚类)
- - Expectation Maximazation(期望最大化)
- Visualization and dimensionality reduction
- - Principle Component Analysis(PCA): 主成分分析
- - Kernel PCA(核函数主成分分析)
- - Locally-Linear Embedding(LLE):局部线性嵌入算法
- - t- distributed Stochastic Neighbor Embedding(t-SNE):t分布随机邻域
- Association rule learning:关联性规则学习
- - Apriori
- - Eclat
2 是否递进式学习(whether or not incrementally learned on the fly)
3 是挖掘模式还是点对点对比(whether detect patterns or compare new data points to known data points)