Boosting is an approach to machine learning based on the idea of creating a highly accurate prediction rule by combining many relatively weak and inaccurate rules [19] and has inspired a lot on theoretical analysis and algorithm design in supervised learning [11, 17].

Motion-based perception can be used in a diverse set of robotics and remote sensing applications. Moreover, it has the unique advantage of being resilient to effect of distance and environmental factors (e.g. The methods proposed are specially useful for long-range perception applications like autonomous driving, perimeter security, or privacy preserving activity monitoring in public or private spaces. In specific, our focus was on predictive accident prevention applications for autonomous driving with the goal of making the roads safer for pedestrians. Like any other remote perception technology, there are also risks involved with misuse of radarbased perception especially in the context of activity monitoring.

Detecting anomalies in real-world multivariate time series data is challenging due to complex temporal dependencies and inter-variable correlations. Recently, reconstruction-based deep models have been widely used to solve the problem. However, these methods still suffer from an over-generalization issue and fail to deliver consistently high performance. To address this issue, we propose the MEMTO, a memory-guided Transformer using a reconstruction-based approach. It is designed to incorporate a novel memory module that can learn the degree to which each memory item should be updated in response to the input data. To stabilize the training procedure, we use a two-phase training paradigm which involves using K-means clustering for initializing memory items. Additionally, we introduce a bi-dimensional deviation-based detection criterion that calculates anomaly scores considering both input space and latent space. We evaluate our proposed method on five real-world datasets from diverse domains, and it achieves an average anomaly detection F1-score of 95.74%, significantly outperforming the previous state-of-the-art methods. We also conduct extensive experiments to empirically validate the effectiveness of our proposed model's key components.