Scott A. Wallace

Game-themed programming assignments: the faculty perspective

We have designed and implemented game-themed programming assignment modules targeted specifically for adoption in existing introductory programming classes. These assignments are self-contained, so that faculty members with no background in graphics or gaming can selectively pick and choose a subset to combine with their own assignments in existing classes. This paper begins with a survey of previous results. Based on this survey, the paper summarizes the important considerations when designing materials for elective adoption. The paper then describes our design, implementation, and assessment efforts. Our result is a road map that guides faculty members in experimenting with game-themed programming assignments by incrementally adopting/customizing suitable materials for their classes.

Developing a community definition and teaching modules for computational thinking: accomplishments and challenges

The NSF-funded Northwest Distributed Computer Science Department (NW-DCSD) project brings together 24 multi-disciplinary faculty from 19 diverse colleges and universities in an effort to change the face of computing in the Pacific Northwest region of the United States. We offer an innovative and inclusive vision of computing in the 21st century and foster opportunities for multi-disciplinary and inter-institutional computing and computer science education collaborations. Over the projects first two years, this community has created 9 engaging, easy-to-use learning modules that teach various levels of computational thinking to two different audiences: non-computer science and computer science undergraduate students. This paper presents the development of a community definition of computational thinking, the learning modules, initial findings, unanticipated challenges, and next steps.

Integrating games and machine learning in the undergraduate computer science classroom

A student will be more likely motivated to pursue a field of study if they encounter relevant and interesting challenges early in their studies. The authors are PIs on two NSF funded course curriculum development projects (CCLI). Each project seeks to provide compelling curricular modules for use in the Computer Science classroom starting as soon as CS 1. In this paper, we describe one curriculum module which is the synergistic result of these two projects. This module provides a series of challenges for undergraduate students by using a game environment to teach machine learning and classic Artificial Intelligence concepts.

Design and evaluation of a Twitter hashtag recommendation system

Twitter has evolved into a powerful communication and information sharing tool used by millions of people around the world to post what is happening now. A hashtag, a keyword prefixed with a hash symbol (#), is a feature in Twitter to organize tweets and facilitate effective search among a massive volume of data. In this paper, we propose an automatic hashtag recommendation system that helps users find new hashtags related to their interests. We propose the Hashtag Frequency-Inverse Hashtag Ubiquity (HF-IHU) ranking scheme, which is a variation of the well-known TF-IDF, that considers hashtag relevancy, as well as data sparseness. Experiments on a large Twitter data set demonstrate that our method successfully yields relevant hashtags for users interest and that recommendations more stable and reliable than ranking tags based on tweet content similarity. Our results show that HF-IHU can achieve over 30% hashtag recall when asked to identify the top 10 relevant hashtags for a particular tweet.

Smart grid line event classification using supervised learning over PMU data streams

Emerging smart grid technology offers the possibility of a more reliable, efficient, and flexible energy infrastructure. A core component of the smart grid are phasor measurement units (PMUs) which offer the ability to capture time-coherent measurements across a geographically distributed area. However, due to the fast sampling rate of these devices, a significant volume of data is generated on a daily basis and this presents challenges for how to leverage the information most effectively. In this paper, we address this challenge by applying machine learning techniques to PMU data for the purpose of detecting line events in a wide-area power grid. Specifically, we use archived synchrophasor data from PMUs located across the Pacific Northwest to train and test a decision tree built using the J48 algorithm. In contrast to other studies exploring machine learning in the context of the smart grid, our work uses PMU data from a large, active, power grid as opposed to data obtained from a simulation. We show that our classifier performs as well as hand-coded rules developed by a domain expert when applied at locations near to a line fault and that it significantly outperforms hand-coded rules when identifying line faults from a distance.

A sensitivity study of PMU-based fault detection on smart grid

Phasor measurement units (PMUs) are widely used in power transmission systems to provide synchronized measurements for the purpose of fault detection. However, how to efficiently deploy those devices across a power grid — so that a comprehensive coverage can be provided at a relatively low cost — remains a challenge. In this paper, we present a sensitivity study of a PMU-based fault detection method using three different distance metrics. This study can serve as a guideline for efficient PMU deployment. To illustrate the effectiveness of this approach, we have derived an alternative PMU placement plan for a power grid. Experimental results show that our PMU placement reduces the required PMU deployment by more than 80% as compared to the original placement, yet still provides similar level of accuracy in fault detection.

Rule-Based Data-Driven Analytics for Wide-Area Fault Detection Using Synchrophasor Data

Synchrophasor technology, also known as wide-area monitoring system technology, utilizes phasor measurement unit (PMU) to monitor real-time system data, which can provide unique insights into the operation of a power grid. In this paper, a rule-based data-driven analytics method for wide-area fault detection in a power system using synchrophasor data is proposed. As a data-driven approach, this method relies on rules created using PMU measurement data, and does not require knowledge of the power systems topology and model. It can detect fault location (bus and line) and fault type for a particular fault event. Three common types of short circuit faults in a power grid, single-line-to-ground, line-to-line, and three-phase faults, can be identified using the proposed method. Fault thresholds used in rules are determined based on theoretical values and recorded PMU data during fault events in Bonneville power administration (BPA)s large power grid. The proposed method is validated by comparing with the recorded field data for fault events provided by BPA. It is found that it can effectively detect most faults with a great accuracy. It has been developed into a software program, and can be readily used by utility companies.

Behavior bounding: an efficient method for high-level behavior comparison

In this paper, we explore methods for comparing agent behavior with human behavior to assist with validation. Our exploration begins by considering a simple method of behavior comparison. Motivated by shortcomings in this initial approach, we introduce behavior bounding, an automated model-based approach for comparing behavior that is inspired, in part, by Mitchells Version Spaces. We show that behavior bounding can be used to compactly represent both human and agent behavior. We argue that relatively low amounts of human effort are required to build, maintain, and use the data structures that underlie behavior bounding, and we provide a theoretical basis for these arguments using notions of PAC Learnability. Next, we show empirical results indicating that this approach is efiective at identifying differences in certain types of behaviors and that it performs well when compared against our initial benchmark methods. Finally, we demonstrate that behavior bounding can produce information that allows developers to identify and fix problems in an agents behavior much more efficiently than standard debugging techniques.

S-assess: a library for behavioral self-assessment

Developing and testing intelligent agents is a complex task that is both time-consuming and costly. This creates the potential that problems in the agents behavior will be realized only after the agent has been put to use. As a result, society is left with a vexing problem: although we can create agents that seem capable of performing useful tasks autonomously, we are simultaneously unwilling to trust these agents because of the inherent incompleteness of testing. In this paper we present a framework that brings validation techniques out of the laboratory and uses them to monitor and constrain an agents behavior concurrent with task execution. Applications of this framework extend well beyond helping to ensure safe agent behavior through run-time validation. They also include the ability to enforce social or environmental policies or to regulate the agents autonomy.

Fast sequence component analysis for attack detection in smart grid

Modern power systems have begun integrating synchrophasor technologies into part of daily operations. Given the amount of solutions offered and the maturity rate of application development it is not a matter of “if” but a matter of “when” in regards to these technologies becoming ubiquitous in control centers around the world. While the benefits are numerous, the functionality of operator-level applications can easily be nullified by injection of deceptive data signals disguised as genuine measurements. Such deceptive action is a common precursor to nefarious, often malicious activity. A correlation coefficient characterization and machine learning methodology are proposed to detect and identify injection of spoofed data signals. The proposed method utilizes statistical relationships intrinsic to power system parameters, which are quantified and presented. Several spoofing schemes have been developed to qualitatively and quantitatively demonstrate detection capabilities.