This repository contains many sub-projects. I have pursued these projects as a means to learn natural language processing. I have approached this field of natural language processing with only deep-learning as the tool. I do not have any linguistics knowledge, and none of that has been used in any of the projects. I believe, and also as has been shown by many deep-learning applications now, that inputting in human knowledge may seem to be better at problem solving for a specific domain, but in the long run letting the achines figure out the patterns for themselves works for the best.
One-Piece is one of the most loved anime series of all times. Greatest story-telling, if you ask me. So, this project ties my love for the great anime and deep-learning together. I feed into the natural language processing algorithm the script for many many one piece episodes. The algorithm learns relationship between a sequence of characters and the following character. Basically it is a multi-class classification problem. We do not have very great results out of this yet. But the project is still on. Here are few snippets generated via natural language processing algorithm.
the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of the sea whth the country which is the cesteen of
i will be able to say that it is the part of the sea whth the straw hat in the straw hat is the part of the sea whth the straw hat with the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat is the part of the sea whth the straw hat with the straw hat is the pirates of the straw hat is the part of the sea whth the straw hat is th
It appears that the algorithm learns a pet phrase, which is not so impressive. But still, seeing that the algorithm can only predict individual characters and gets most of the spelling right is a bit nice. I believe more training can help in this scenario. Also, future prospcts for this include training on words rather than characters. Currently, the project uses LSTMs. In future, we can explore things like Conv1D and other like algorithms too.
This project is not entirely natural language processing. It is a sequence prediction project which uses the same techniques as a natural-language-processing tasks. We have used Dense Networks (MLPs), GRUs and LSTMs to predict temperature of next day, given temperature of a whole day back. The baseline was taken as the same temperature that was previous day at the same time. The deep-learning techniques successfully crossed this baseline.
All these projects are types of text-classification. reuters is a many-class classification, whereas spam-classification and imdb-sentiment analysis are binary text-classification projects. Again, they use similar techniques but to a different effect.