John Reeder

Fast parallel outlier detection for categorical datasets using MapReduce

Outlier detection has received considerable attention in many applications, such as detecting network attacks or credit card fraud The massive datasets currently available for mining in some of these outlier detection applications require large parallel systems, and consequently parallelizable outlier detection methods. Most existing outlier detection methods assume that all of the attributes of a dataset are numerical, usually have a quadratic time complexity with respect to the number of points in the dataset, and quite often they require multiple dataset scans. In this paper, we propose a fast parallel outlier detection strategy based on the Attribute Value Frequency (AVF) approach, a high-speed, scalable outlier detection method for categorical data that is inherently easy to parallelize. Our proposed solution, MR-AVF, is based on the MapReduce paradigm for parallel programming, which offers load balancing and fault tolerance. MR-AVF is particularly simple to develop and it is shown to be highly scalable with respect to the number of cluster nodes.

Human-Computer Collaboration in Adaptive Supervisory Control and Function Allocation of Autonomous System Teams

The foundation for a collaborative, man-machine system for adaptive performance of tasks in a multiple, heterogeneous unmanned system teaming environment is discussed. An autonomics system is proposed to monitor missions and overall system attributes, including those of the operator, autonomy, states of the world, and the mission. These variables are compared within a model of the global system, and strategies that re-allocate tasks can be executed based on a mission-health perspective (such as relieving an overloaded user by taking over incoming tasks). Operators still have control over the allocation via a task manager, which also provides a function allocation interface, and accomplishes an initial attempt at transparency. We plan to learn about configurations of function allocation from human-in-the-loop experiments, using machine learning and operator feedback. Integrating autonomics, machine learning, and operator feedback is expected to improve collaboration, transparency, and human-machine performance.

Interactively evolved modular neural networks for game agent control

As the realism in games continues to increase, through improvements in graphics and 3D engines, more focus is placed on the behavior of the simulated agents that inhabit the simulated worlds. The agents in modern video games must become more life-like in order to seem to belong in the environments they are portrayed in. Many modern artificial intelligence approaches achieve a high level of realism but this is accomplished through significant developer time spent scripting the behaviors of the non-playable characters or NPCs. These agents will behave in a believable fashion in the scenarios they have been programmed for, but do not have the ability to adapt to new situations. In this paper we introduce a modularized, real-time evolution training technique to evolve adaptable agents with life-like behaviors. Online performance during evolution is also improved by using selection mechanisms found in temporal difference learning methods to appropriately balance the exploration and exploitation of control policies. These methods are implemented and tested using the XNA framework producing very promising results regarding efficiency of techniques, and demonstrating many potential avenues for further research.

Command and control of teams of autonomous systems

The command and control of teams of autonomous vehicles provides a strong model of the control of cyber-physical systems in general. Using the definition of command and control for military systems, we can recognize the requirements for the operational control of many systems and see some of the problems that must be resolved. Among these problems are the need to distinguish between aberrant behaviors and optimal but quirky behaviors so that the human commander can determine if the behaviors conform to standards and align with mission goals. Similarly the commander must able to recognize when goals will not be met in order to reapportion assets available to the system. Robustness in the face of a highly variable environment can be met through machine learning, but must be done in a way that the tactics employed are recognizable as correct. Finally, because cyber-physical systems will involve decisions that must be made at great speed, we consider the use of the Rainbow framework for autonomics to provide rapid but robust command and control at pace.

Team Search Tactics Through Multi-Agent HyperNEAT

User defined tactics for teams of unmanned systems can be brittle and difficult to define. The state and action space grows with each new system added to the team which increases the difficultly in designing robust behaviors. In this paper we present a method for using Multi-agent HyperNEAT to develop tactics for a team of simulated unmanned systems that is robust to novel situations, and scales with the number of team members. We focus on the tactics of a search area coverage task, where the need for team work, and robust asset management are critical to success.

Human interactive machine learning for trust in teams of autonomous robots

Unmanned systems are increasing in number, while their manning requirements remain the same. To decrease manpower demands, machine learning techniques and autonomy are gaining traction and visibility. One barrier is human perception and understanding of autonomy. Machine learning techniques can result in “black box” algorithms that may yield high fitness, but poor comprehension by operators. However, Interactive Machine Learning (IML), a method to incorporate human input over the course of algorithm development by using neuro-evolutionary machine-learning techniques, may offer a solution. IML is evaluated here for its impact on developing autonomous team behaviors in an area search task. Initial findings show that IML-generated search plans were chosen over plans generated using a non-interactive ML technique, even though the participants trusted them slightly less. Further, participants discriminated each of the two types of plans from each other with a high degree of accuracy, suggesting the IML approach imparts behavioral characteristics into algorithms, making them more recognizable. Together the results lay the foundation for exploring how to team humans successfully with ML behavior.

Convolution neural networks for ship type recognition

Algorithms to automatically recognize ship type from satellite imagery are desired for numerous maritime applications. This task is difficult, and example imagery accurately labeled with ship type is hard to obtain. Convolutional neural networks (CNNs) have shown promise in image recognition settings, but many of these applications rely on the availability of thousands of example images for training. This work attempts to under- stand for which types of ship recognition tasks CNNs might be well suited. We report the results of baseline experiments applying a CNN to several ship type classification tasks, and discuss many of the considerations that must be made in approaching this problem.

Haptic enhancements for collaborative scenarios in virtual environment

The more senses a virtual reality system can simulate, the greater the realism and sense of immersion experienced by the user. Nearly every system provides the vital visual sense, and most provide some degree of audio reproduction as well. However, when the task involves a significant amount of manual manipulation of virtual objects, or when the scenario calls for both non-verbal and non-visual communication between its participants, the sense of touch becomes vital.

Evolving cost functions for model predictive control of multi-agent UAV combat swarms

Recent advances in sampling-based Model Predictive Control (MPC) methods have enabled the control of nonlinear stochastic dynamical systems with complex and non-smooth cost functions. However, the main drawback of these methods is that they can be myopic with respect to high-level tasks, since MPC relies on predicting dynamics within a short time horizon. Furthermore, designing cost functions which capture high-level information may be prohibitive for complex tasks, especially multi-agent scenarios. Here we propose a hierarchical approach to this problem where the NeuroEvolution of Augmenting Topologies (NEAT) algorithm is used to build cost functions for an MPC trajectory optimization algorithm known as Model-Predictive Path Integral (MPPI) control. MPPI and NEAT are particularly well-suited to one another since MPPI can control an agent in a way that minimizes a non-differentiable cost function (including logic or non-smooth functions), while NEAT can build a neural network comprised of any arbitrary activation functions, including those which are non-differentiable or logic-based. We utilize this approach in controlling agile swarms of unmanned aerial vehicles (UAVs) in a simulated swarm vs. swarm combat scenario.

Amplifying human ability through autonomics and machine learning in IMPACT

Amplifying human ability for controlling complex environments featuring autonomous units can be aided by learned models of human and system performance. In developing a command and control system that allows a small number of people to control a large number of autonomous teams, we employ an autonomics framework to manage the networks that represent mission plans and the networks that are composed of human controllers and their autonomous assistants. Machine learning allows us to build models of human and system performance useful for monitoring plans and managing human attention and task loads. Machine learning also aids in the development of tactics that human supervisors can successfully monitor through the command and control system.