ML-Overview.txt

            					Machine Learning - Algorithm
            
1. Linear Regression --> y = mx+c ---> Multilinear regression,least square regression..., ----> weather forecasting,stock,sales predictions...,

2. Logistic Regression --> sigmoid/logit function --> Bi-classifaction/multi-classification ---> sentiment analysis.

3. SVM --> curve fitting & margin calculation --> Bi-classifaction ---> male/female.

4. K-means --> Ecludean distance and mean --> Clustering ---> customer grouping.

5. K-NN --> Ecludean distance --> clustering & classification ---> customer grouping/ object recognition.

6. Decision Tree --> Entropy and gini index --> regression & classification ---> anything limited to dataset.

7. Random Forest --> ensemble method --> regression & classification ---> iris flower classification.

8. AdaBoost --> ensemble method --> regression & classification ---> XGBoost,Gradient Boosting ----> iris flower classification.
 
9. LDA --> word vs sentence matrix in vector --> Topic modeling ---> LSI ----> Generating topics for paragraph.

10. TF-IDF  --> word count & number of sentence --> Topic modeling and important words identification ---> Fetching keywords from the paragraph.

11. Naives bayes --> Conditional Probability --> regression & classification ---> decision making based on some parameters.

12. Apriori --> Combination and number of occurences --> assosative identification ---> super market discount products.

13. Eclat --> Combination and number of occurences --> assosative identification ---> identification of super market discount products.

14. Hierarichy --> Ecludean distance --> clustering ---> object grouping.

                                                   Neural Network
                                                   
1. Perceptron --> weights and threshold --> decision making based on parameters.

2. Sigmoid --> weights and bias in sigmoid function --> decision making based on parameters with small changes in the input.

3. Neural Network --> based on activation function in hidden layer --> MNIST Digits classification (multiple process of same input to learn better).

4. Backpropagation --> based on error function --> weight updates from the back.

5. Gradient descent --> based on error function and learning rate ---> stochostic gradient descent ----> weight updates from the back.


							Deep Learning
						
1. Basic terminalogy --> optimizer, epoch, dropout, object function, softmax, ReLu, Dense, validation/training accuracy, verbose.

2. Basic Architecture --> based on input,hidden, output layer --> MNIST Digits classification
						
3. CNN --> convolution, shared weights ,pooling --> Cifar image classifcation.

4. word embedding --> word2vec, GLovE --> converting word to vectors.

5. Vanila RNN --> based on past output --> character based sentence prediction.

6. LSTM --> input, forget & output gate --> Bi-DAF,DCN,DMN ---> Sentiment analysis 

7.GRU --> update and reset gate --> sentiment analysis.


							Projects
							
1. Image --> Product recommendation based on facial features and emotions.
         --> Object recognition and identification.
         --> Image Description
         
2. Text  --> Question formation
         --> Inference generation 
         
3. Numerical --> Electricity usage prediction (LSTM/GRU)