Scribd
Upload
15 views9 pages

CODTECH Internship Presentation Project Based

The internship at CODTECH IT SOLUTIONS focuses on a project for SMS spam detection using a hybrid model that combines CNN, LSTM, and TF-IDF techniques. The project aims to develop a robust and scalable solution, achieving over 98% accuracy on the test dataset. Key learnings include practical skills in NLP and deep learning, as well as experience in the end-to-end machine learning workflow.

Uploaded by

kilaruharika2003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
15 views9 pages

CODTECH Internship Presentation Project Based

The internship at CODTECH IT SOLUTIONS focuses on a project for SMS spam detection using a hybrid model that combines CNN, LSTM, and TF-IDF techniques. The project aims to develop a robust and scalable solution, achieving over 98% accuracy on the test dataset. Key learnings include practical skills in NLP and deep learning, as well as experience in the end-to-end machine learning workflow.

Uploaded by

kilaruharika2003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Internship at CODTECH IT

SOLUTIONS
Domain: Machine Learning
Project: SMS Spam Detection Using
Feature Fusion of CNN, LSTM, and TF-
IDF
Presented by: Kilaru Harika
About the Internship
• Organization: CODTECH IT SOLUTIONS,
Hyderabad
• Mode: Online
• Domain: Machine Learning
• Duration: Jan 5, 2025 – May 5, 2025
• Focus: Real-world project on SMS Spam
Detection
Project Overview
• Title: SMS Spam Detection Using Feature
Fusion of CNN, LSTM, and TF-IDF
• Goal: To accurately detect spam messages in
SMS using a hybrid model
• Techniques: Deep learning and classical NLP
techniques
• Outcome: Developed a robust and scalable
model
Objectives of the Project
• Analyze limitations of existing spam detection
methods
• Implement multi-type feature extraction
(CNN, LSTM, TF-IDF)
• Fuse features into a single vector
• Apply feature selection techniques
• Train and evaluate classification model
• Develop a scalable and deployable solution
Methodology
• 1. Preprocessing: Remove punctuation,
lowercase text, stop-word removal,
tokenization
• 2. Feature Extraction: CNN for local patterns,
LSTM for sequence learning, TF-IDF for term
weighting
• 3. Feature Fusion: Combine outputs into a
unified feature vector
• 4. Feature Selection: Remove redundant
features
Implementation Highlights
• Used Word2Vec/GloVe for word embeddings
• 1D CNN and Bi-LSTM layers built using Keras
• TF-IDF vectors computed and combined with
deep features
• Final dense neural network layer trained with
binary cross-entropy loss
Results and Metrics
• Evaluation Metrics: Accuracy, Precision, Recall,
F1-score
• Achieved >98% accuracy on test dataset
• Hybrid feature fusion model outperformed
standalone models
• Error analysis highlighted challenges with
short ambiguous messages
Key Learnings
• Developed practical skills in NLP and deep
learning
• Improved understanding of hybrid model
design
• Learned to handle real-world text
classification challenges
• Experienced end-to-end ML workflow from
data cleaning to evaluation
Conclusion
• This internship provided hands-on experience
in Machine Learning
• The project demonstrates the power of
feature fusion in text classification
• Model ready for integration into real-time
SMS filtering systems
• Thank you! Open to questions.

You might also like