ART-EMAP: A Neural Network Architecture for Learning and Prediction by Evidence Accumulation

This paper introduces ART-EMAP, a neural architecture that uses spatial and temporal evidence accumulation to extend the capabilities of fuzzy ARTMAP. ARTEMAP combines supervised and unsupervised learning and a medium-term memory process to accomplish stable pattern category recognition in a noisy input environment. The ART-EMAP system features (i) distributed pattern registration at a view category field; (ii) a decision criterion for mapping between view and object categories which can delay categorization of ambiguous objects and trigger an evidence accumulation process when faced with a low confidence prediction; (iii) a process that accumulates evidence at a medium-term memory (MTM) field; and (iv) an unsupervised learning algorithm to fine-tune performance after a limited initial period of supervised network training. ART-EMAP dynamics are illustrated with a benchmark simulation example. Applications include 3-D object recognition from a series of ambiguous 2-D views. ART-EMAP: An ARTMAP System for 3-D Object Recognition ART-EMAP (Fig. 1) is a neural network architecture that extends fuzzy AllTMAP (Carpenter, Grossberg, Markuzon, Reynolds, and Rosen, 1992) to accomplish target object or pattern class recognition in noisy or ambiguous input environments. During performance, ART-EMAP integrates spatial evidence distributed across coded recognition categories to predict a pattern class. When a decision criterion determines the pattern class choice to be ambiguous, additional input from the same unknown class is sought. Evidence from multiple inputs accumulates until the decision criterion is satisfied and a high confidence prediction can be made. Accumulated evidence can also be used by the predictive mapping to fine-tune the system during unsupervised rehearsal learning. ART-EMAP was developed to address the problem of 3-D object recognition by 2-D view recognition. Applications would include a vision system capable of sampling different 1Supported in part by British Petroleum (89-A-1204), DARPA (AFOSR 90-0083, and ONR N0001492-J-4015), the National Science Foundation (NSF IRI 90-00530), and the Office of Naval Research (ONR N00014-9 1-J -4100). 'Supported in part by the Air Force Office of Scientific Research (AFOSR 90-0083), British Petroleum (89-A-1204), the National Science Foundation (NSF IRI 90-00530), and the Oflice of Naval Research (ONR NOOO 14-9 1-J-4100). fuzzy ART a ART EMAP Architecture