ShreyasP123/Anomaly-detection-in-time-series-data

Anomaly Detection in Time Series Data

Model Description

This model is designed for Anomaly Detection in Time Series Data using a simple feedforward neural network. It takes a time series dataset as input and identifies anomalies based on learned patterns. The model is implemented in PyTorch and trained using synthetic data.

Model Architecture

• Input Layer: 10-dimensional feature vector
• Hidden Layer: 128 neurons, ReLU activation
• Output Layer: 1 neuron, Sigmoid activation (for binary classification)

Training Details

• Dataset: Synthetic time series data (100 samples, 10 features each)
• Loss Function: Binary Cross Entropy Loss (BCELoss)
• Optimizer: Adam (learning rate = 0.001)
• Epochs: 10

How to Use

Load the Model

import torch
import torch.nn as nn

# Define model architecture
class AnomalyDetectionModel(nn.Module):
    def __init__(self, input_dim=10):
        super(AnomalyDetectionModel, self).__init__()
        self.fc = nn.Sequential(
            nn.Linear(input_dim, 128),
            nn.ReLU(),
            nn.Linear(128, 1),
            nn.Sigmoid()
        )
    
    def forward(self, x):
        return self.fc(x)

# Load the model
model = AnomalyDetectionModel()
model.load_state_dict(torch.load("anomaly_detection_model.pth"))
model.eval()

Make Predictions

# Dummy input sample (10 features)
input_data = torch.rand(1, 10)
prediction = model(input_data).item()

if prediction > 0.5:
    print("Anomaly detected!")
else:
    print("No anomaly detected.")

Applications

• Fraud detection in financial transactions
• Network intrusion detection
• Predictive maintenance in IoT systems
• Fault detection in industrial equipment

Limitations

• The model was trained on a synthetic dataset; real-world datasets may require retraining.
• Hyperparameter tuning and feature engineering can improve performance.

License

MIT License

Author

ShreyasP123