Efficient Late Fusion for Indonesian Sarcasm Detection
Combining an optimized classical baseline with transformer predictions for Indonesian sarcasm detection, while examining the limits of the results.
Read the storyThe projects, the process, and everything learned along the way.
Combining an optimized classical baseline with transformer predictions for Indonesian sarcasm detection, while examining the limits of the results.
Read the storyDigitizing village administration through document archives, regulatory workflows, and automated database backups.
Read the storyAutomating meeting minutes and connecting recordings, attendance, invitations, and email distribution in one workflow.
Read the storyComparing approaches to six-month food price forecasts and examining why low error alone does not guarantee useful predictions.
Read the storyDeveloping practical labs, video walkthroughs, and visual explanations for a machine learning learning path accessed by over 80 learners.
Read the storyExploring continual learning for multilabel emotion classification while reducing catastrophic forgetting.
Read the storyComparing hourly electricity demand forecasts and investigating why strong next-hour accuracy did not translate into reliable 30-day predictions.
Read the storyExploring which image regions influence a classifier and why a correct prediction can still depend on surprising contextual clues.
Read the storyBuilding a desktop application to detect drowsiness through eye closure, yawning, and blink patterns.
Read the storyUnderstanding HTTP by building a Java socket server with threaded requests, file serving, protected routes, and a desktop interface.
Read the storyOrganizing tutorial content into ordered presentation steps, with a shareable page and PDF export.
Read the story