Researchers have developed a fuzzy selective state model to forecast consumer preferences, utilizing a Mamba-type architecture. This model describes customer preferences as triangular fuzzy numbers, with their evolution governed by a selective state space model. The model's parameters are computed from crisp features, ensuring that the fuzzy state remains a proper triangular number and a convex combination of past observations, thus preventing the accumulation of forecast vagueness over time. A synthetic example demonstrated that this lightweight model achieves a lower mean absolute error compared to other forecasting methods and offers linguistically interpretable estimates of forecast vagueness, capable of detecting customer switches to competitors. AI
IMPACT Introduces a novel fuzzy logic approach integrated with Mamba architecture for more interpretable and precise consumer preference forecasting.
RANK_REASON The cluster contains a research paper published on arXiv detailing a new modeling approach. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
