Fruit Classiffier with Dimensionality Reduction
Fruits-360 Classifier with only 188 dimensions using PCA. Detecting Outliers with Isolation Forest.
View the Project on GitHub alxmares/Fruits-360-with-Dim-Reduction-only-188-dims
This project leverages dimensionality reduction techniques to classify fruit images from the Fruits-360 dataset, reducing dimensions from 30,000 (100x100x3) to just 188 while maintaining high accuracy.
🛠️ Tools and Technologies Used
🧠 Algorithms Used
| Algorithm | Description |
|---|---|
| Isolation Forest | Anomaly detection to identify outliers and improve model robustness. |
| K-Nearest Neighbors (KNN) | Used for its simplicity and effectiveness in classification tasks. |
| Support Vector Machine (SVM) | Chosen for its ability to handle high-dimensional spaces effectively. |
| Multilayer Perceptron (MLP) | Implemented for its deep learning capabilities, providing flexibility and precision. |
| Hard Voting | Ensemble method that aggregates model predictions for improved accuracy. |
| Soft Voting | Ensemble method that uses probability weights to refine predictions. |
| Principal Component Analysis (PCA) | Dimensionality reduction technique to reduce features while preserving variance. |
| t-Distributed Stochastic Neighbor Embedding (t-SNE) | Visualization technique for representing high-dimensional data in lower dimensions. |
🔽 Dimensionality Reduction
Principal Component Analysis (PCA)
To tackle the challenge of high-dimensional data, Principal Component Analysis (PCA) was employed to reduce the number of dimensions from 30,000 (representing each pixel in a 100x100x3 image) to just 188. This substantial reduction not only decreases computational complexity but also helps prevent overfitting, enhancing the model’s generalization ability.
- Original Dimensions: 30,000 (100x100x3)
- Reduced Dimensions: 188
- Purpose: Minimize data redundancy while preserving as much variability as possible, ensuring the essential features of the images are retained.
📊 Key Results
| Modelo | Tipo de clasificación | Accuracy |
|---|---|---|
| Hard voting | Hard voting | 90.30% |
| Soft Voting | Soft Voting | 89.94% |
| Support Vector Machine | Probabilística | 89.95% |
| K-Nearest Neighbors | Probabilística | 88.39% |
| Multi-layer Perceptron | Probabilística | 86.19% |
Visualization
Anomaly detection using Isolation Forest
