Alexander A. Petrov

Taylor W. Webb, Zachary Dulberg, Steven M. Frankland et al.

Extrapolation -- the ability to make inferences that go beyond the scope of one's experiences -- is a hallmark of human intelligence. By contrast, the generalization exhibited by contemporary neural network algorithms is largely limited to interpolation between data points in their training corpora. In this paper, we consider the challenge of learning representations that support extrapolation. We introduce a novel visual analogy benchmark that allows the graded evaluation of extrapolation as a function of distance from the convex domain defined by the training data. We also introduce a simple technique, temporal context normalization, that encourages representations that emphasize the relations between objects. We find that this technique enables a significant improvement in the ability to extrapolate, considerably outperforming a number of competitive techniques.

Vsevolod Khithov, Alexander Petrov, Igor Tishchenko et al.

Autonomous fixed-wing UAV landing based on differential GPS is now a mainstream providing reliable and precise landing. But the task still remains challenging when GPS availability is limited like for military UAVs. We discuss a solution of this problem based on computer vision and dot markings along stationary or makeshift runway. We focus our attempts on using infrared beacons along with narrow-band filter as promising way to mark any makeshift runway and utilize particle filtering to fuse both IMU and visual data. We believe that unlike many other vision-based methods this solution is capable of tracking UAV position up to engines stop. System overview, algorithm description and it's evaluation on synthesized sequence along real recorded trajectory is presented.