AI Foundations

The AI Foundations course provides a comprehensive introduction to core machine learning and deep learning concepts, spanning supervised and unsupervised learning, neural network architectures, computer vision, and natural language processing. It progresses from classical algorithms to modern deep learning frameworks and state-of-the-art models. The course also introduces Generative AI, equipping students with both theoretical understanding and hands-on experience using industry-standard libraries.

Objectives of the Course

1. Build a strong foundation in machine learning by understanding and applying core supervised learning algorithms and unsupervised learning techniques along with appropriate model evaluation metrics.
2. Develop proficiency in deep learning architectures by designing and training core neural network models and landmark architectures using industry-standard frameworks and libraries.
3. Gain practical exposure to NLP by mastering text preprocessing, representation techniques, and transformer-based models, laying a strong foundation for real-world AI application development.

Layout of Course

Supervised Learning

Classification algorithms: Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Decision Trees (ID3, C4.5, CART), Naive Bayes.

Regression algorithms: Linear Regression, Polynomial Regression, Lasso/Ridge Regression.

Ensemble Methods: Random Forest, Bagging, Boosting - AdaBoost, XGBoost, CatBoost.

Model Evaluation Metrics: Classification metrics - Confusion Matrix, Accuracy, Precision, Recall (Sensitivity), F1-Score, ROC-AUC.

Libraries: NumPy, scikit-learn, pandas, Matplotlib, Seaborn

Unsupervised Learning

Clustering algorithms: K-Means Clustering, Hierarchical Clustering, DBSCAN, Gaussian Mixture Models (GMM).

Dimensionality reduction algorithms: Principal Component Analysis (PCA), t-SNE.

Model Evaluation Metrics: Regression Metrics - Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), R-Squared (R²).

Libraries: NumPy, scikit-learn, pandas, Matplotlib, Seaborn

Deep Neural Networks

Neural Networks: Artificial Neural Networks (ANN), Backpropagation, Single-layer and Multi-layer Perceptrons (MLP), Feedforward and feedback structures, Input / Hidden / Output layers, Weights, Biases, Activation functions (Sigmoid, ReLU, Tanh), Dropout for regularization, Batch Normalization, Local Response Normalization, Tuning Hyper-parameters.

Deep Learning Frameworks & Libraries: PyTorch, TensorFlow, Keras

Core Deep Learning Frameworks

Convolutional Neural Networks (CNN): Convolutional layers (filters, kernels, stride, padding), Pooling layers (Max / Average pooling), Flattening, Fully Connected (FC) layers, Softmax output.

Classic CNN Architectures: LeNet-5, AlexNet, VGGNet (VGG-16/VGG-19), GoogLeNet / Inception, ResNet (Residual Networks), MobileNet.

Sequential Models: Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU).

Deep Learning Frameworks & Libraries: PyTorch, TensorFlow, Keras, fast.ai, Huggingface Transformers

Computer Vision Models

Single-Stage Object Detection Models: YOLO (v8/v9/v10/v11), SSD (Single Shot Multibox Detector), and RetinaNet.

Two-Stage Object Detection Models: R-CNN (Regions with CNN), Fast R-CNN, Faster R-CNN.

Semantic Segmentation Models: U-Net, DeepLabv3+, PSPNet (Pyramid Scene Parsing Network), SegNet, SAM2 (Segment Anything Model 2), Mask2Former.

Frameworks, Libraries & AI Model Platforms: Ultralytics (YOLO), MMDetection, Detectron2, OpenCV, Huggingface

Natural Language Processing (NLP)

Natural Language Processing (NLP): Origins, Challenges, and Applications of NLP, Text tokenization, Normalization, and Language modeling.

Preprocessing: Stemming, lemmatization, stop-word removal, and morphological analysis.

Text Representation: Bag-of-words, TF-IDF, and word embeddings (Word2Vec, GloVe).

Syntactic Analysis: Part-of-Speech (POS) tagging, constituency parsing, and dependency parsing.

Semantics and Information Extraction: Named Entity Recognition (NER), distributional semantics, and topic modeling (LDA, NMF).

Models: Bag of Words (BoW), BERT (Bidirectional Encoder Representations from Transformers), GPT (Generative Pre-trained Transformer), ELMo (Embeddings from Language Models), Transformer-XL & RoBERTa.

Libraries & Frameworks: Hugging Face Transformers, spaCy, NLTK (Natural Language Toolkit), Gensim, TextBlob.