NeuralLib: Deep Learning Models for Biosignals Processing
Collection
Collection of pre-trained models for biosignals processing. These models are optimized for easy integration with the NeuralLib package (pip). • 2 items • Updated
Model Card for ECGPeakDetector
Collection: NeuralLib: Deep Learning Models for Biosignals Processing
Description: GRU-based model for ECG peak detection.
- Architecture: GRUseq2seq
- Model Name: ECGPeakDetector
- Task: peak detection
- Train Dataset: private_gib01
Biosignal(s): ECG
Sampling frequency: 360
Benchmark Results
Validation Loss: 0.1488
Training Time: 11375.49 seconds
FLOPs per timestep: 0
Number of trainable parameters: 0
Hyperparameters
| Parameter | Value |
|---|---|
| bidirectional | True |
| dropout | 0 |
| hid_dim | [32, 64, 64] |
| learning_rate | 0.001 |
| model_name | ECGPeakDetector |
| multi_label | True |
| n_features | 1 |
| n_layers | 3 |
| num_classes | 1 |
| task | classification |
Example
import NeuralLib.model_hub as mh
model_name = ECGPeakDetector()
model = mh.ProductionModel(model_name=model_name)
signal = torch.rand(1, 100, 1) # Example input signal
predictions = model.predict(signal)
print(predictions)
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support