CMU SoS Lablet Quarterly Executive Summary_July 2018

A. Fundamental Research
High level report of result or partial result that helped move security science forward-- In most cases it should point to a "hard problem". These are the most important research accomplishments of the Lablet in the previous quarter.

Metrics. In response to the growing number of demonstrations of the fragility of machine learned classification networks, metrics have been proposed to define concepts of "similarity" among inputs that, despite similarity are classified differently. Such metrics would have great value in addressing the challenges of adversarial inputs. A recent paper from the CMU lablet team (CV-COPS workshop) highlights the difficulty of this task through demonstrations that many of the proposed metrics are, in fact, themselves fragile and unsuitable for the intended purpose.

Humans and resilience. Explanation generation techniques are being developed to enable human operators to understand outcomes from the multi-objective probabilistic planning that underlies some resiliency architectures. The explanations are in the form of natural language descriptions appropriate to human operators.

B. Community Engagement(s)
Research interaction in the community including workshops, seminars, competitions, etc.

C. Educational Advances
Impact to courses or curriculum at your school or elsewhere that indicates an increased training or rigor in security research.