Classify different latex equations into one of the four classes which are "Unknown, Simple, Fraction, Linear Equation For Degree One". Almost 20k latex equations were collected and then trained a keras based lstm model using that dataset. The final test accuracy was 98%.
Model Training.ipynb is being used for training
use "mapping.pickle" and "prediction.ipynb" for making predictions for new equation files, with the trained model provided as "weights-improvement-29-0.98.h5"