M-3 - Architectural Enhancements #33
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Owner
Thomasbehan
commented
Mar 31, 2024
Thomasbehan
commented
- AI Model Downloading: Implemented functionality for automatic downloading of AI models, facilitating ease of setup.
- Service Layer Improvements: The service layer has been enhanced for increased efficiency and maintainability.
- Operational Commands Directory: Introduced a commands directory for operational scripts, streamlining execution processes.
- Directory Cleanup: Removed outdated core and ai folders to maintain a clean and organized codebase.
… Categorization - Implemented iterative pagination to fetch all available images from the ISIC Archive, ensuring comprehensive data collection. - Added functionality to categorize images into benign and malignant based on metadata, facilitating more structured data analysis. - Introduced train-test split (80-20) for categorized images, laying the groundwork for robust model training and evaluation. - Implemented efficient handling of previously downloaded images and graceful skipping of failed downloads, optimizing the scraping process and ensuring resilience against interruptions. - Minor code optimizations and refactoring for better readability and maintainability. This update significantly improves the utility of the DataScraper, making it a more powerful tool for gathering and organizing dermatological image data for machine learning purposes.
…ing no data in training for the validator
This commit introduces the `hparam_tuning.py` module, designed to enhance model performance by systematically exploring a range of hyperparameters. This new module integrates seamlessly with TensorBoard for comprehensive tracking and analysis of each tuning session, facilitating data-driven decision-making in model development.
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…o best-performing trial settings 🏆
…and session reuse
… SkinCancerDetector tests
…tecture diagram 🎨
…ncerDetector - Added class weights calculation to balance training data. - Integrated data augmentation directly in preprocessing. - Included F1 score as a metric for better performance evaluation. - Simplified data pipeline by embedding rescaling into the model.
…ncerDetector - Added class weights calculation to balance training data. - Integrated data augmentation directly in preprocessing. - Included F1 score as a metric for better performance evaluation. - Simplified data pipeline by embedding rescaling into the model.
…nced model performance - Updated ReduceLROnPlateau to monitor 'val_auc', focusing on separation capabilities in imbalanced data. - Adjusted ModelCheckpoint to save models with the highest 'val_f1_score', prioritizing balance between precision and recall. - Shifted EarlyStopping to monitor 'val_f1_score', ensuring early halt on significant metric improvement stagnation.
…nced model performance - Updated ReduceLROnPlateau to monitor 'val_auc', focusing on separation capabilities in imbalanced data. - Adjusted ModelCheckpoint to save models with the highest 'val_f1_score', prioritizing balance between precision and recall. - Shifted EarlyStopping to monitor 'val_f1_score', ensuring early halt on significant metric improvement stagnation.
…nced model performance - Updated ReduceLROnPlateau to monitor 'val_auc', focusing on separation capabilities in imbalanced data. - Adjusted ModelCheckpoint to save models with the highest 'val_f1_score', prioritizing balance between precision and recall. - Shifted EarlyStopping to monitor 'val_f1_score', ensuring early halt on significant metric improvement stagnation.
…ayer enhancements, and operational commands; removed old 'core' and 'ai' folders
… than a usable model
… detection performance
… detection performance
… detection performance
… detection performance
… detection performance
… detection performance
Reordered evaluation metrics to better reflect diagnostic importance: prioritized Recall and AUC for their critical role in minimizing missed diagnoses and balancing sensitivity/specificity. F1 Score, Precision, Accuracy, and Binary Accuracy follow, in order of decreasing impact on patient outcomes. This change aims to enhance model evaluation aligning with medical priorities. 🚀🔍
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.