Weightless neuro-symbolic GPS trajectory classification

A bag of nodes primer on weightless graph classification

“…Plot Binarizationthe initial strategy used in [11], consists of plotting R 2 points in a canvas and using the plot image as a binary input vector.…”

Analysis of road travel behaviour based on big trajectory data

“…In addition, the OD data extracted from the massive taxi‐hailing data sets can fully reflect residents’ travel rules and travel demand distributions [6]. A comprehensive understanding of road travel behaviour can be used for many aspects, such as path planning [7–10], traffic states estimation [11, 12], and location‐based social network [13, 14].…”

Process Mining Classification with a Weightless Neural Network

“…The WiSARD weightless neural network name comes from its authors (Wilkie, Stonham, and Aleksander's Recognition Device) [9] and it was initially created to recognize images as a hardware architecture [10]. Although WiSARD has been previously categorized as a supervised learning method, and new developments reveal its usage as unsupervised learning as well [11].…”

“…Figure 1 illustrates this dynamic and exemplifies how WiSARD operates on the training phase. [11] The network learns from example writing a '1' in all memory addresses, associated to a particular input. Input mapping in the classification phase follow the same process as in the training examples.…”

“…The highest score designates the output class. [11] The standard WiSARD model may present some drawbacks. One of them occurs when it handles very noisy data.…”