Application of Multi-class Supervised Algorithms to Identify COVID-19 Viral Strains
summary
Applies DFT plus Shannon entropy to create a single magtropy feature from viral genomes, trains tree-based models (XGBoost, decision trees) with 10-fold CV and a 70/30 split on datasets of 50 sequences per class, then tests on a 200-sequence SARS‑CoV‑2 holdout. Reports 97% genus accuracy but only 33% at the realm level, finds one‑vs‑rest better here than one‑vs‑one for four classes, and notes ECOC / loss‑based decoding remain untested due to limited data.
AI-written summary · about
chapters (11)
- 0:00Introduction and overview
- 0:45Multi-class classification basics
- 2:52Binary reduction strategies
- 9:17Error correcting output codes
- 13:06Loss-based decoding approach
- 17:40Viral taxonomy classification
- 20:05Feature engineering methods
- 23:11Preliminary results summary
- 27:23One vs rest vs one vs one
- 29:08ECoc and loss-based future plans
- 31:13Conclusions and acknowledgments
related talks
28mMachine Learning vs The Flu
22mCOVID-19: Data Science & Expertise
56mA Global Comparison of COVID-19 Variant Waves and Relationships with Clinical and...
31mAn AI Approach to Malware Similarity Analysis: Mapping the Malware Genome With a Deep Neural Network
57mHow Machine Learning and AI Can Support the Fight Against COVID-19
30mData Driven Security
27mAutomated Discovery of Cancer Types from Genes
40mDemystifying File Similarity for Malware Detection