Delve into neural networks, backpropagation, optimizers, and build deep learning models using PyTorch

• What is a Neural Network
• Forward and backward propagation
• Activation Functions
• Gradient Descent
• Vanishing and exploding
• Weights & Bias
• Regularization:

- Dropout

• Batch vs Mini Batch vs Stochastic:

- Batch Normalisation

• Optimizers:

- GD with Momentum

- RMSProp

- Adam

• Autoencoders:

- Anomaly Detection

• PyTorch

<Mini Project> Deep Learning model building and training using PyTorch for a classification Problem

Work with text data using NLP techniques, vectorization methods, and build sequence models like RNNs and LSTMs

• What is NLP?
• Various methods to transform textual data into numerical data:

- Bag of Words, TF-IDF, etc.

- Word2vec, doc2vec, etc.

• Data Preprocessing in NLP
• RNN, LSTM, GRU

<Mini Project> A sentiment classification model in PyTorch

Understand transformers, BERT, GPT, prompt engineering, RAG systems, and fine-tuning large language models

• Attention & Transformers.
• BERT:

- <Mini Project>

• Generative Models:

- How does GPT-2 work?

- How do decoder models predict the sentences? Different decoding methods:

    ->Greedy search

    ->Breedy search

    ->Sampling

    ->Tok-k sampling

    ->Top-p sampling.

• Sentence Transformer:

- What is semantic search ?

- What is vector db?

- KNN Brute Force vs Approx Nearest Neighbor(ANN) concept in Information retrieval (IR) for a RAG system?

    ->IVF

    ->LSH

    ->HNSW

• Prompt Engineering:

- Best practices when writing a prompt

- Zero-shot vs few-shot

- What is chain of thoughts ?

• Langchain, RAG

- What is memory buffer in langchain for chat models

- Models and different tasks supported by langchain

- <Mini Project> Build a RAG model for Q/A

Fine Tuning of LLMs

Apply your skills by developing and presenting a complete, real-world data science project from scratch

Guided Learning:

Comprehensive documents, videos, and assignments will be provided for the following topics, with ample time allocated based on your semester schedule.

1. Python Fundamentals - Self Learning (Build a strong foundation in Python basics, including data structures, functions, and decorators)

Data Structures: List, Tuples, Sets, Dictionary

Functions & argument variations, decorators

1. SQL Advanced - Self Learning (Master advanced SQL concepts and learn to integrate SQL queries within Python using libraries like Pandasql and DuckDB)

Schema

Joins

GroupBy

Index

Window Functions

Subquery

Case when

Connecting & Querying from Python

Use SQL in python on pandas:

- Pandasql

- duckdb

1. Math Foundations for Data Science - Self Learning (Gain essential mathematical knowledge in calculus, linear algebra, probability, and statistics for data science applications.)

Calculus

Linear Algebra

Probability

Statistics

1. Advanced DSA using Python (Explore key advanced data structures such as stacks, queues, and linked lists to optimize problem-solving skills.)

Stack

Queue

Linked List