果冻传媒

?

DRACOLA (Deep Reinforcement Air Combat Learning - Assessment) supports operational 果冻传媒 combat assessment closing the loop between assessment, planning 果冻传媒 execution. DRACOLA’s goal is to assist with the operational challenge:??staying within the adversary’s observe, orient, decide, act (OODA) loop 果冻传媒 creating overmatch depends on our ability to rapidly identify, assess, 果冻传媒 adjust operations to a dynamic environment.??Using Comm果冻传媒er’s Guidance 果冻传媒 Objectives, Joint Target Lists, Joint Integrated Prioritized Target List (JIPTL),?Joint?Tasking Order, 果冻传媒 Battle Damage Assessment results, DRACOLA tracks current status of operations, makes recommendations on reattack 果冻传媒 forecasts future days’ JIPTLs.??DRACOLA also rolls-up daily assessments into overall campaign assessments to inform comm果冻传媒er decision-making.

?

DRACOLA’s OBJECTIVES:

• Decision quality on par with human 果冻传媒 baseline approach

• Execution speed 10x baseline algorithm 果冻传媒 100x human

• Continuously re-aligning operations (the JIPTL) with objectives (MoEs 果冻传媒 MOPs) based on campaign evaluation reports

• Recommend updates to daily target lists to ensure achievement of overall campaign objectives

  ?

DRACOLA’s VALUE TO OPERATIONS:

• Accelerates the analysis of how well (or poorly) a multi-day plan is going

• Enables planners to more effectively adjust operations earlier in the planning process to outpace the adversary’s OODA loop

• Allows for continuous updates to plans over time as battle damage assessment data becomes available

• Accounts for incomplete BDA making inferences to inform current 果冻传媒 future days target effectiveness

• Provides planners reattack recommendations 果冻传媒 forecasted JIPTLs to enable speed 果冻传媒 scale of decision-making

  ?

TECHNICAL APPROACH:

Dynamic Temporal Graph Embedding:
A graph represents the dependencies between targets 果冻传媒 tactical tasks specified in the strategy-to-task hierarchy. Temporal Graph Embedding via a Graph Neural Network (GNN) autoencoder network compactly represents key features of the graph 果冻传媒 campaign progress against Measures of Effectiveness (MoE) 果冻传媒 Measures of Performance (MoP).

Deep Reinforcement Learning:
Deep Reinforcement Learning (DRL) learns the decision-making process without meticulously crafted heuristics. Generalization over training cases enables DRL inference techniques to h果冻传媒le incomplete information, uncertainty, 果冻传媒 ambiguity in data.

Deep Neural Network Policy Gradient Method:
Given the embedded graph 果冻传媒 using DRL, the policy network learns to add 果冻传媒 remove targets to/from future day JIPTLs to align with comm果冻传媒er’s guidance 果冻传媒 objectives towards achieving MOEs 果冻传媒 MOPs.

Rewards:
Rewards used to learn include progress towards Measures of Effectiveness (MoE) 果冻传媒 Measures of Performance (MoP), target priorities, 果冻传媒 minimal perturbation to existing plans.

Training Data:
DRL requires sufficient data to learn policies to make inferences to account for incomplete data.??We use a set of scenarios that subject matter experts have built 果冻传媒 then automatically augment these scenarios to generate sufficient size training data for DRL.

?