Skip to content

An api which predicts if a text is spam or not from a model which is trained using LSTM, a special kind of recurrent neural network.

Notifications You must be signed in to change notification settings

RohanKaran/spam-classifier

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

FastAPI ReactJS API API

Spam Classifier

Demo

View the live app here

About

Datasets:

Dataset 1 and Dataset 2 are collected from UCI Machine Learning Repository.

Model:

  • A specific architecture of a neural network is created, still a "blank slate" in terms of what it "knows". Its core structure is that of a LSTM (long-short-term-memory), a specific kind of recurrent neural network with some clever modifications aimed at enhancing its ability to "remember" things between non-adjacent locations in a sequence, such as two displaced positions in a string of text;
  • The network is trained: that means it will progressively adapt its internal (many thousands of) parameters in order to best reproduce the input training set. Each individual neuron in the network is a relatively simple component - the "intelligence" coming from their sheer quantity and the particular choice of parameters determining which neurons affect which other and by how much;
  • Once the training process has finished, the script carefully saves everything (model, tokenizer and associated metadata) in a format that can be later loaded by the backend in a stand-alone way.

Web Application Details:

About

An api which predicts if a text is spam or not from a model which is trained using LSTM, a special kind of recurrent neural network.

Topics

Resources

Stars

Watchers

Forks