Files and Solutions

1. Q1 - Rain probability

   • Q1.py - Python implementation
   • Q1.cpp - C++ implementation

2. Q2 - Phonemas

   • Q2.py - Python implementation
   • Q2.cpp - C++ implementation

3. Q3 - Most frequent words

   • Q3.cpp - C++ implementation

4. Q4 - Connectionist temporal classification

   • CTC/ - The implementation of Q4 as a custom loss function for PyTorch, supporting both forward and backward passes. The CTC forward-backward algorithm was used to calculate the probability and the loss function as well as its gradient. The CTC forward algorithm was implemented based on the functions of this repository and the backward algorithm was similarly implemented using dynamic programing. For testing, cd to /CTC.