Artificial Intelligence (AI) : Training Course

• Why/What machine learning
• Real time applications of machine learning
• Data science project life cycle
• Overview of data, tools and processes
• Environment setup – Anaconda distribution setup

• Basic data structures – List, Set, Tuple, Dictionary
• Functions and loops
• Object-oriented programming: classes &objects
• Packages
• Overview of python libraries (pandas, skelarn, keras, spacy,nltk,etc)
• Directory and File management
• Exception handling

• Data types
• Measures of central tendency – mean, median and mode
• Measures of variability and spread – range, IQR and standard deviation, boxplot, skewness and kurtosis

Detailed course contents:

• Random variable and probability
• Probability rules and types
• Bayes’ theorem
• Histogram and Probability density function
• Probability distributions
• Bernoulli, Geometric, Binomial, Poisson and exponential distributions

• Normal distribution
• Z-score
• Central limit theorem
• Confidence intervals
• Hypothesis testing – p value, Null and alternate hypothesis
• Type I and Type II Errors
• T-test for paired and unpaired data
• Chi-Square test – Goodness of fit and Independence
• F distribution - ANOVA

• Basic numpy and scipy operations
• Pandas library and its functions

• Data collection
• Analyzing data using pandas
• Visualization techniques – matplotlib, seaborn and plotly
• Scatter plot, bar plot, box plot, pie chart, heat map etc.

• Data cleaning and transformations
• Missing value handling
• Outlier handling
• Train/Valid/test split – random and stratified
• Feature engineering

• Supervised learning– Regression, Classification
• Unsupervised learning– Clustering, dimensionality reduction
• Reinforcement learning

• Simple linear regression
• Assumptions of linear regression
• Multiple Linear Regressions
• Polynomial Regression
• Evaluation- MSE, RMSE, correlation, R Squared, Adjusted RSquared
• Bias- variance tradeoff
• Use case of Linear regression

• Sigmoid function
• Odds ratio and logit
• Logistic regression
• Evaluation - Confusion matrix, AUC-ROC
• K-fold cross validation
• Use case of logistic regression

• K-means clustering
• Hierarchical clustering
• Evaluation

• Decision trees
• Ensemble models
• Bagging – Random forest
• Boosting – XGBoost, LightGBM, Catboost
• Support vector machines SVM
• Naïve bayes
• KNN – K nearest neighbors
• Evaluating all models with one use case

• Train, Test &Validation Set
• Cross-validation (K-fold CV)
• Hyper-parameterization
• Randomized search
• Stacking

• PCA- principle component analysis
• Matrix factorization
• SVD
• Use case of dimensionality reduction

• Neural network architecture
• Neuron and activation functions
• Loss function
• Back propagation and gradient descent
• Use case of neural networks

• Deep learning architecture
• Parameters and hyper parameters
• Hyper parameter tuning
• Stochastic gradient descent
• Mini batch
• Optimiser-RMSProp,ADAM,ADAGRAD
• Overfitting solutions
• Regularization techniques-Lasso,ridge and elastic
• Dropout
• Early stopping and data augmentation
• Cross entropy and Softmax
• Use case of deep learning

• RNN - Recurrent neural networks architecture
• RNN variants
• Vanishing gradient and Exploding gradient problem
• LSTM Long short-term memory architecture and working
• GRU Gated recurrent unit architecture and working
• Use cases of RNN, LSTM and GRU

• Computer vision tasks and challenges
• Convolution network advantages
• CNN architecture – kernel/Filter, activation, padding andpooling
• Use case of CNN

• Autoencoders
• Transfer learning
• GAN- Generative adversarial network