load_dataset.py

import json
import pandas as pd
from datasets import Dataset, DatasetDict

def load_bci_fps_dataset(data_dir):
    """
    Load BCI-FPS dataset for Hugging Face.
    
    Args:
        data_dir (str): Path to dataset directory
    
    Returns:
        DatasetDict: Hugging Face dataset
    """
    # Load neural data
    neural_data = []
    with open(f"{data_dir}/neural_data.jsonl", 'r') as f:
        for line in f:
            if line.strip():
                neural_data.append(json.loads(line))
    
    # Load intent stream
    intent_stream = []
    with open(f"{data_dir}/intent_stream.jsonl", 'r') as f:
        for line in f:
            if line.strip():
                intent_stream.append(json.loads(line))
    
    # Create datasets
    datasets = {
        "neural_data": Dataset.from_list(neural_data),
        "intent_stream": Dataset.from_list(intent_stream)
    }
    
    # Load handwriting samples if exists
    try:
        with open(f"{data_dir}/handwriting_samples.json", 'r') as f:
            handwriting = json.load(f)
        datasets["handwriting"] = Dataset.from_list(handwriting)
    except:
        pass
    
    # Load metadata
    with open(f"{data_dir}/metadata.json", 'r') as f:
        metadata = json.load(f)
    
    dataset_dict = DatasetDict(datasets)
    dataset_dict.info.metadata = metadata
    
    return dataset_dict

# Example usage for Neuralink research
if __name__ == "__main__":
    dataset = load_bci_fps_dataset("./bci_data")
    
    print(f"Dataset keys: {list(dataset.keys())}")
    print(f"Neural data samples: {len(dataset['neural_data'])}")
    print(f"Intent stream samples: {len(dataset['intent_stream'])}")
    
    # Example: Extract motor imagery trials
    motor_trials = [d for d in dataset['neural_data'] if d.get('type') == 'motor_imagery_trial']
    print(f"Motor imagery trials: {len(motor_trials)}")