PEARLNN (Parameter Extraction And Reverse Learning Neural Network) is an open-source, community-powered AI tool that automatically extracts electronic component parameters from measurement data or datasheet images.

• CSV files with measured waveform data (oscilloscope exports, simulation results)
• Images of waveforms from datasheets, manuals, or publications
• Initial conditions and component specifications
• Multiple signal types: voltage, current, switching, frequency response

• Accurate parameter values for electronic components
• Uncertainty estimates for each parameter (Bayesian confidence intervals)
• Iterative improvement with each usage
• Validation metrics showing fit quality

• MOSFETs: Rds_on, Ciss, Coss, Crss, Qg, Vth, switching times
• BJTs: Beta, Vce_sat, Vbe_on, capacitances, Ft
• Op-Amps: GBW, slew rate, Vos, Ib, CMRR, PSRR
• Capacitors: ESR, ESL, leakage, dissipation factor, temperature coefficient
• Inductors/Transformers: DCR, Q-factor, saturation current, SRF
• Diodes: Vf, reverse recovery, junction capacitance, leakage
• Voltage Regulators: line/load regulation, dropout voltage, PSRR

# Dual-mode neural network
- **Backpropagation Network**: Main pattern recognition from waveforms
- **Bayesian Layers**: Uncertainty quantification and confidence scoring
- **Multi-modal Input Processing**: Handles both CSV numerical data and image waveforms
- **Transfer Learning**: Builds on previous community knowledge

• Supervised Learning: Uses known parameter→waveform relationships
• Community Training: Each user's successful fits improve the shared model
• Incremental Learning: Models get smarter with more usage
• Uncertainty-Aware: Knows when it's confident vs. guessing

# Distributed model storage (completely free)
- **GitHub Gists**: For model metadata and version tracking
- **IPFS Network**: For actual model weights and training data
- **Multiple Fallbacks**: Ensures always-available access
- **Anonymous Contributions**: No login required for sharing

1. User installs: pip install pearlnn
2. First use: Downloads latest community model automatically
3. Analysis: Processes local CSV/image, extracts parameters
4. Training: If user provides validation, model improves locally
5. Sharing: Improved model uploaded to community pool
6. Everyone benefits: Next user gets smarter model

Raw Input → Feature Extraction → Neural Network → Parameter Output
    ↑              ↑                  ↑               ↑
 CSV/Image   Waveform Analysis   Bayesian NN     Uncertainty
                                  + Backprop      Estimates

• Automated waveform feature extraction from images/datasheets
• Bayesian uncertainty tells users when to trust results
• Incremental learning without catastrophic forgetting
• Model compression for easy sharing
• Cross-platform command-line tool
• Batch processing for multiple components

# Analyze MOSFET from oscilloscope CSV
pearlnn extract mosfet --csv gate_switching.csv --vds 24V

# Extract op-amp parameters from datasheet image  
pearlnn extract opamp --image gain_plot.png --supply ±15V

# Batch process multiple measurements
pearlnn batch-analyze --folder lab_measurements/ --component capacitor

# Get uncertainty estimates
pearlnn analyze --uncertainty diode --csv reverse_recovery.csv

# Check community model status
pearlnn community --status

# Share your improvements
pearlnn share --contribution

# Download latest knowledge
pearlnn community --sync

• Free alternative to $50,000+ commercial tools
• AI-powered accuracy that improves over time
• Community knowledge from thousands of users
• Command-line automation for batch processing

• Reproducible parameter extraction
• Uncertainty quantification
• Collaborative model development
• Growing knowledge base

• Learn component behavior through AI analysis
• Professional-grade tools for free
• Contribute to community knowledge
• Practical parameter fitting for projects

• MOSFET parameter extraction from CSV/Images
• Basic Bayesian neural network
• Community model sharing via IPFS
• Command-line interface

• Support for 10+ component types
• Advanced uncertainty quantification
• Web interface for non-programmers
• Model validation suite

• Industry-standard accuracy
• Real-time parameter estimation
• Integration with SPICE simulators
• Mobile app for field measurements

PEARLNN democratizes electronics characterization by making advanced parameter extraction accessible to everyone, while creating a continuously improving community knowledge base that benefits all users with each analysis performed.

- "pylint.args": ["--disable=C0111","--disable=E1101", "--max-line-length=120"] as pylint arguments for ignoring issues with pylint compatibility with opencv-python

- use firebase for hosting the shared model file instead of ipfs