Privacy-Preserving Gesture Recognition with Explainable Type-2 Fuzzy Logic Based Systems

Smart homes are a growing market in need of privacy preserving sensors paired with explainable, interpretable and reliable control systems. The recent boom in Artificial Intelligence (AI) has seen an ever-growing persistence to incorporate it in all spheres of human life including the household. This growth in AI has been met with reciprocal concern for the privacy impacts and reluctance to introduce sensors, such as cameras, into homes. This concern has led to research of sensors not traditionally found in households, mainly short range radar. There has been also increasing awareness of AI transparency and explainability. Traditional AI black box models are not trusted, despite boasting high accuracy scores, due to the inability to understand what the decisions were based on. Interval Type-2 Fuzzy Logic offers a powerful alternative, achieving close to black box levels of performance while remaining completely interpretable. This paper presents a privacy preserving short range radar sensor coupled with an Explainable AI system employing a Big Bang Big Crunch (BB-BC) Interval Type-2 Fuzzy Logic System (FLS) to classify gestures performed in an indoor environment.