Finding the Path

Finding the path together.

Are you a mentor? Do you have one? Do you think you need one? I recently attended a joint meeting of the Section of Medical Students, Section of Residents, and First Five Years in Practice Committee. When asked who had a mentor, most attendees raised their hands. This was a surprise to many in attendance. The Canadian Association of Internes and Residents 2013 National Resident Survey indicated...

Path-Finding in Networks

Algorithms that find connections between concepts in semantic networks are inherently inefficient due to their dependence on piecemeal representation of potential paths. A method of reorganizing a link-based network into a path-based representation is argued to allow more efficient search techniques. A path-based representation is presented in terms of the knowledge representation language Conc...

Mode Directed Path Finding

Learning from multi-relational domains has gained increasing attention over the past few years. Inductive logic programming (ILP) systems, which often rely on hill-climbing heuristics in learning first-order concepts, have been a dominating force in the area of multi-relational concept learning. However, hill-climbing heuristics are susceptible to local maxima and plateaus. In this paper, we sh...

Grid-Based Path-Finding

Differentiable Path Finding

Our motivation is to scale value iteration to larger environments without a huge increase in computational demand, and fix the problems inherent to Value Iteration Networks (VIN) such as spatial invariance and unstable optimization. We show that VINs, and even extended VINs which improve some of their shortcomings, are empirically difficult to optimize, exhibiting instability during training an...