Automatic target recognition and a related problem of non-cooperative identification friend or foe often require fusing of multiple sensor information into a unified battlefield picture. State of the art approaches to this problem attempt to solve it in steps: first targets are detected, then target tracks are estimated, these are used to correlate, or associate information between multiple sensors, the associated information is combined into the unified picture and targets are identified. A drawback of dividing the problem into smaller steps is that only partial information is utilized at every step. For example, detection of a target in clutter may not be possible on a single frame of a single sensor, and the target motion information may have to be utilized requiring track estimation, so several steps have to be performed concurrently. This paper describes such a concurrent solution of the multiple aspects of this problem based on the MLANS neural network that utilizes internal world models. The internal models in MLANS encode a large number of neural weights in terms of relatively few model parameters so MLANS learning occurs with significantly fewer examples than required with unstructured neural networks. We describe the model-based neural network paradigm, MLANS, present results on MLANS concurrently performing detection and tracking, adaptively estimating background properties, and learning and classifying similar U.S. and foreign military vehicles. The MLANS performance is compared to that of the multiple hypothesis tracker, the classical statistical quadratic classifier, and the nearest neighbor classifier, demonstrating significant performance improvement.