Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

MA74

observable steps called a ‘kill chain.’ Using burgeoning techniques within analytic systems and machine learning, data produced from early kill chain steps can be used to mitigate downstream consequences. An application of ensemble classification using supervised learning techniques is explored. 2 - Bots in Nets: Empirical Analysis of Bot Evidence in Online Social Networks Ross Schuchard, George Mason University, 4400 University Drive, Fairfax, VA, 22030, United States Online social networks (OSNs) continue to play an increasing role as a primary source of information in today’s society. The emergence of social bots within OSNs to diffuse information at scale has given rise to many efforts to detect bots. While methodologies employed to detect the evolving sophistication of bots continue to improve, much work can be done to characterize the impact of bots on actual communication networks. This study proposes a social network analysis framework to characterize the pervasiveness and relative importance of bots in various OSN conversations. 3 - Fake News: Fearing the End of Truth, a Quantitative Risk Analysis Travis Trammell, Stanford University, Stanford,, CA, United States Strategically using information to affect the views of a population is certainly nothing new and dates all the way back to the earliest development of political systems. The speed of distribution and the number of people that can be reached by leveraging the modern information infrastructure is unprecedented. The rapid distribution of fake news can cause contagion, manipulate markets, spark conflict, or fracture strategic relations. Probabilistic Risk Analysis (PRA) can be leveraged to quantitatively describe the risk associated with fake news by examining all relevant factors to evaluate associated probabilities and costs resulting from fake news campaigns. 4 - Saving the Army from Cyberspace: A Strategy to Seize the Initiative Using Big Data Gregory Bew, U.S. Army Cyber Command, Fort Belvoir, VA, United States In the information age the defense of the Department of Defense Information Network (DoDIN) is arguably our most mission essential task. Mission command, logistics, medical support, and weapons systems all rely on access to the DoDIN or we will operate in a severely degraded state. In war, that means people will die. To ensure access, Army Cyber Command decided to rely on big data and developed a phased approach to seize initiative, increase confidence, and change our culture to defend our cyber terrain. The results of this approach have created the DoD’s largest and most capable cyber big data platform, providing a common environment to develop apps and analytics to help the Army defend the DoDIN. 5 - An Advanced Analytics Framework for Optimizing Army Cyber Mission Force Readiness and Manning Nathaniel D. Bastian, Operations Research Scientist, Army Cyber Given the unique expertise required of military personnel to execute the DoD cyber mission, the US Army created the Cyber Branch to establish managed career fields for Army cyber warriors, while providing a force structure with successive opportunities for career development and talent management via leadership and broadening positions, technical training, and advanced education. In order to optimize Cyber Mission Force readiness and manning levels across the Army’s operating and generating forces, we proffer the Cyber Force Manning Model (CFMM) to project the optimal number of accessions, promotions and personnel inventory for each cyber specialty across the Army cyber enterprise. n MA76 West Bldg 212C Production Logistics and Supply Chain General Session Chair: Feng Ju, Arizona State University, Tempe, AZ, 85281, United States Co-Chair: Feifan Wang, Arizona State University, Arizona State University, Tempe, AZ, 85281, United States 1 - A Facility Location Problem with Inventory-level-dependent Hazard Zones Mina Aliakbari, Texas A&M, College Station, TX, United States, Joseph Geunes We consider a location problem for administrative facilities and hazardous stock. Each stock location creates an inventory-level-dependent hazard zone within which no administrative facilities may be located. Increasing the stock at a location increases the radius in which administrative offices are forbidden. We seek a set of administrative and stock locations that can accommodate required warehouse and office space at a minimum total cost. Such problems arise, for example, on military bases or in the production of hazardous materials. This results in a large-scale combinatorial optimization problem, for which we propose using Lagrangian relaxation and heuristic solution techniques. Institute, U.S. Military Academy, West Point, NY, 10996, United States, Andrew O. Hall, Christopher B. Fisher

n MA74 West Bldg 212A Joint Session MCDM/Practice Curated: Multiobjective Optimization: Theory and Applications Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Margaret M. Wiecek, Clemson University, Clemson, SC, 29634- 1907, United States 1 - On Highly Robust Efficient Solutions for Uncertain Multiobjective Linear Programs Margaret M. Wiecek, Clemson University, Mathematical Sciences Dept, Clemson, SC, 29634-1907, United States We develop properties of the highly robust efficient (HRE) solutions to uncertain multiobjective linear programs (UMOLPs) with objective-wise uncertainty in the objective function coefficients. A characterization using the cone of improving directions, several bound sets on the HRE set, and a robust counterpart for a class of UMOLPs are provided. A bilevel method for computing the HRE solutions is proposed. 2 - Multiobjective Design of a Fin in a Steady-State Regime Lakmali Weerasena, University of Tennessee Chattanooga, 615 McCallie Ave, Chattanooga, TN, 37403, United States, Boris Belinskiy, James Hiestandzand. Removal of waste heat to another material or the environment by convection and radiation is important in everyday life and industrial applications. Extended surfaces are often used to remove heat and such surface extensions for convective heat transfer frequently are called fins. The design of a fin is modeled as bi- objective optimization problem. The efficiency of the fin and its mass are considered as two objective functions and the multi-objective optimization carried out to maximize the efficiency and the minimize the mass simultaneously. The approach is based on a piece-wise constant design of the fin. 3 - Balancing Mission Goals and Maintenance Demands using a Force Structure Model Chris Grubb, Systems Planning and Analysis, Inc., Alexandria, VA, United States, Stephanie Diane Brown, Jonathon Leverenz, Brian Chen A Force Structure Model designs a day-to-day schedule for a set of assets to meet mission goals and maintenance demands while satisfying travel, duration, and operational constraints. Tension in the schedule arises from the desire to maximize time allotted for maintenance against the need to provide a sufficient number of assets for missions each day. A network-based model and branch-and- bound algorithm are used to investigate tradeoffs between allocating time for mission goals and maintenance demands. The network model offers more flexibility than existing timetabling methods when designing the schedule and the algorithm is shown to outperform CPLEX for longer, more difficult problems. 4 - Multi-objective Optimization for Political Districting with Explicit Fairness Considerations Rahul Swamy, Champaign, IL, 61820, United States, Douglas M. King, Sheldon H. Jacobson Political redistricting is a multi-objective problem with conflicting objectives such as compactness, population balance, etc. While the problem is well-studied, the use of political fairness metrics has been relatively under-explored. In addition, contiguity enforcement within an exact method has been a challenging task. This research presents a multi-objective approach explicitly considering political criteria such as efficiency gap and competitiveness within a branch and cut framework. The results show that compactness does not always ensure political fairness, and vice versa. Joint Session MAS/Practice Curated: Advanced Analytics for Military Cyber Security, Defense and Readiness Joint Session Chair: Nathaniel D. Bastian, PhD, Army Cyber Institute, West Point, NY, 10996, United States 1 - Early Warning Systems for Cyber Security Isaac Faber, PhD Candidate, Stanford University, 141 Aryshire Farn Ln #113, Stanford, CA, 94305, United States Through the paradigm of early warning systems, from risk analysis, early-stage cyber attack signals can be generated using machine learning techniques. The past ten years have seen the growth of interest in cybersecurity. As cyber threats become more sophisticated, system defenders must keep pace with better methods of detection and response. Advanced cyber-attacks are a set of discrete, n MA75 West Bldg 212B

146