Marvel Collin / Research
Quantization Effects on Explainability in Indonesian Transformer-Based Sentiment Analysis
Quantization Effects on Explainability in Indonesian Transformer-Based Sentiment Analysis is a 2026 research paper by Helena Aurelia Sanjaya, Marvel Collin, Bertrand Geraldo Tjahyadi and Prof. Derwin Suhartono.
Research on applying model quantization techniques to sentiment analysis tasks. Includes multi-seed evaluation, explainable AI (XAI) analysis, statistical testing, and NusaX benchmark evaluation. Investigates how quantized models perform compared to full-precision baselines on Indonesian language sentiment data. Published in Procedia Computer Science (ICCSCI 2026).
Topics: NLP, Quantization, Sentiment Analysis, XAI, ICCSCI