It is important to introduce computer programming concepts to K-12 students in preparation and motivation for college-level programming courses. Furthermore, generalization and abstraction are skills needed in computer science and other STEM majors. The Collaborative Partnership to teach mathematical Reasoning through Computer PRogramming (CPR2) Instructional Model (IM) was developed to use Python programming as a vehicle for explicit instruction in mathematical generalization and abstraction. Students are introduced to programming for the purpose of exploring mathematical concepts. They write code embedded with mathematical general expressions, observe the behavior of those expressions at scale in the execution of their programs, make conjectures about emerging patterns, and write convincing arguments to support these conjectures. During a two-week summer institute in 2020, middle school math teachers participated in professional development using the CPR2 instructional model. Teachers were provided with an introduction to programming and then shown how to apply the instructional model to math concepts while addressing computer science and math standards in the process. During the following fall and spring, the CPR2-trained teachers piloted lessons using Python programming to explore math concepts in their own classrooms. In this paper, details about the instructional model, the collaborative design research process used to refine professional development sessions, the key role of experienced teacher mentors in supporting new participants, and the incorporation of programming into the teachers' math classrooms are detailed. The results of design sessions and the pilot implementation are also discussed.

Students who are majoring in computer-related disciplines should have a reasonable understanding of data representation. Concepts like encoding, range and precision, overflow, round-off error, and data compression are all vitally important. Unfortunately, students often struggle to recognize the practicality of these concepts. To address this concern, we created a 6-module lesson plan that uses real-world examples to help students connect with this material. When incorporated into our Computer Science I (CS1) course, we found measurable improvements in both student perceptions and associated computational skills.

It is a widely held belief among computer science researchers that studying computer science enhances critical thinking, problem solving, and creativity; and it should be a part of the K-12 classrooms. Our home state has joined other states in mandating the latter. Issues of curriculum development, teacher preparation, pre-service and in-service training, funding and logistics, are all important. Over several years, we have developed a novel approach to using computer programming to explicitly teach mathematical generalization and abstraction. In this study, we applied this instructional model to pre-service elementary teachers (PSETs). PSETs participated in four days of explicit instruction where students wrote mini programs designed to push them towards generalization of statistics concepts found in the elementary classroom. Along with instruction, PSETs were given information regarding the important contributions of computer science and abstraction in their future elementary classrooms. Results showed that PSETs did show improvement in conceptual understanding and also showed that some of the teachers were initially unconvinced that the curriculum was relevant to them or to their future classrooms. The instruction improved PSET attitude and showed that training to meet new computer science (CS) standards in K-6 can and should address this reluctance.

The financial industry has been a frequent and heavy target of cyberattacks.This trend is likely to continue as the cybersecurity threat remains high in the financial sector.Through a critical analysis of current risks, potential business strategies, and software and hardware strategies, a set of best practices is presented that will help prevent and mitigate cyberattacks for financial institutions.These guidelines should be used as a practical application for financial organizations and can also serve as a basis for future research.

Using the methods of bibliometric analysis and content analysis, we collected and analyzed relevant literature on game-based learning and assessment systems.We explored the key elements, assessment contents, assessment methods, and the data mining process of game-based educational assessment.Finally, we proposed a framework of game-based learning and assessment systems.This framework emphasizes the universality of game-based learning and assessment systems and highlights the important role of an interaction behavior log in mining the learning data.This framework can not only be used for educational games but can also be extended to learning and assessment systems based on a simulation environment.

The advent of storing images on cloud platforms has introduced serious privacy concerns. The images are routinely scanned by machine learning algorithms to determine the contents. Usually the scanning is for marketing purposes but more malevolent purposes include criminal activity and government surveillance. The images are automatically analysed by machine learning algorithms. Notably, deep convolutional neural networks perform very well at identifying image classes. Obviously, the images could be encrypted before storing to cloud platforms and then decrypted after downloading. This would certainly obfuscate the images. However, many users prefer to be able to peruse the images on the cloud platform. This creates a difficult problem in which users prefer images stored in a way so that a human can understand them but machine learning algorithms cannot. This paper proposes a novel technique, termed seam doppelganger, for formatting images using seam carving to identify seams for replacement. The approach degrades typical image classification performance in order to provide privacy while leaving the image human-understandable. Furthermore, the technique can be largely reversed providing a reasonable facsimile of the original image. Using the ImageNet database for birds, we show how the approach degrades a state-of-the-art residual network (ResNet50) for various amounts of seam replacements.

The aircraft cargo industry still maintains vast amounts of the maintenance history of aircraft components in electronic (i.e. scanned) but unsearchable images. For a given supplier, there can be hundreds of thousands of image documents only some of which contain useful information. Using supervised machine learning techniques has been shown to be effective in recognising these documents for further information extraction. A well known deficiency of supervised learning approaches is that annotating sufficient documents to create an effective model requires valuable human effort. This paper first shows how to obtain a representative sample from a supplier's corpus. Given this sample of unlabelled documents an active learning approach is used to select which documents to annotate first using a normalised certainty measure derived from a soft classifier's prediction distribution. Finally the accuracy of various selection approaches using this certainty measure are compared along each iteration of the active learning cycle. The experiments show that a greedy selection method using the uncertainty measure can significantly reduce the number of annotations required for a certain accuracy. The results provide valuable information for users and more generally illustrate an effective deployment of a machine learning application.

With the pervasiveness of the necessity of computational thinking across fields, more universities are requiring courses that build students' ability to think computationally. Computer Science 1 (CS1) is one such course where computational thinking is required This paper summarizes the work of two CS faculty who co-taught separate sections of CS1 for five semesters. Course modifications were made to augment CS1 with support inside and outside of the classroom for students to be successful. The use of in class tutors and requiring design documents were two of the primary modifications made to the course. A variety of data was collected in areas such as student planning, program design, frustration, and resources used to determine what relationships impact student success. One of the main benefits observed was an increase in the student pass rate.

AbstractResearch into risky decision making (RDM) has become a multidisciplinary effort. Conversations cut across fields such as psychology, economics, insurance, and marketing. This broad interest highlights the necessity for collaborative investigation of RDM to understand and manipulate the situations within which it manifests. A holistic understanding of RDM has been impeded by the independent development of diverse RDM research methodologies across different fields. There is no software specific to RDM that combines paradigms and analytical tools based on recent developments in high-performance computing technologies. This paper presents a toolkit called RDMTk, developed specifically for the study of risky decision making. RDMTk provides a free environment that can be used to manage globally-based experiments while fostering collaborative research. The incorporation of machine learning and high-performance computing (HPC) technologies in the toolkit further open additional possibilities such as scalable algorithms and big data problems arising from global scale experiments.

Computer science and math education researchers have long believed that a symbiotic relationship exists between their disciplines [7]. In fact, in its early days, computer science education programs were often co-located in a math department. Stenger et al. [13] developed an instructional treatment that uses computer programming as an explicit method for teaching abstraction and generalization in the STEM classroom. This instructional strategy uses computer programming to explore the essential characteristics of a mathematical concept and to push learners to advance in levels of abstraction. In this study, results are shown from a professional learning session using computer programming activities, mathematical arguments, and programming on an S2 robot to push middle and high school computer science, math, and science teachers (N=25) to improve their level of generalization over area expansion of a triangle with respect to the expansion of the sides of the triangle. The programming activities served as a laboratory to expose and explain what happened in the minds of learners as they explored and learned to generalize this geometry concept. The researchers used an initial genetic decomposition to evaluate the learner's level of abstraction. Follow up interviews were conducted with 6 participants. The analysis, using APOS as a framework, categorized mathematical behaviors at the Action, Process or Object level. The data demonstrated how computer programming activities influenced teachers' mental images and pushed them to higher levels of abstraction.

Java is arguably today's most popular and widely used object-oriented programming language. Learning Java is a daunting task for students, and teaching it is a challenging undertaking for instructors. To assess students' object-oriented programming skills with Java, we developed the "Department-Employee" project. In this article, we review the history of object-oriented programming, provide an overview of object-oriented programming with Java, and present a summary of existing Java projects and their limitations. We also provide the project specifications as well as the course background, grading rubric, and score reports. Survey data are presented on students' backgrounds, as well as students' perceptions regarding the project. Results from the instructor score reports, correlation of the project score and the final course score, and student perceptions show that the "Department-Employee" project is effective in assessing students' object-oriented programming skills with Java.

With the proliferation of wireless networks and mobile computing devices, providing location-aware technology and services to new applications has become important for developers. Our main contribution is an efficient location discovery algorithm that bounds the localization error. Providing an efficient localization technique is critical in resource-constrained environments that include mobile devices and wireless networked sensors. Applicable in centralized and distributed environments, our algorithm, based on finding the smallest circle enclosing the intersection of n disks, runs in O(n 2) time. We then extend our work to the problem of finding the smallest disk that includes the set of points common to n disks and excluded from the interiors of m other disks. Finally, we show performance results from the implementation of our algorithms in which, under some conditions, localization estimates for 500 nodes in a 500x500 ft region can be found with a mean error of one foot and a two-foot error bound.

With the proliferation of wireless sensor networks, providing location-aware technology and services to new applications have become important for developers. Localization is the problem of determining the positions of nodes in an ad hoc network. With the constrained resources of network sensors, providing robust localization services remains a fundamental research challenge facing the entire sensor network development community. The initial localization problem that we addressed was to design and develop a working system that could locate equipment, such as a laptop or video projector. Ferret, the localization system developed, uses two different ranging techniques to help locate an object to within one meter. Our next goal was to identify the locations of all nodes in a sensor network given the locations of a small subset of nodes. The system we developed, LESS or Localization using Evolution Strategies in Sensornets, provides substantial energy savings over existing techniques while providing comparable accuracy. We then introduce an efficient location discovery algorithm that bounds the localization error. Our algorithm, based on finding the smallest circle enclosing the intersection of n disks, runs in O( n2) time. We extend our work to the problem of finding the smallest disk that includes the set of points common to n disks and excluded from the interiors of m other disks. Many localization techniques say that a node can either be localized or it cannot. We present a location discovery algorithm that provides, for every node in the network, a position estimate, as well as an associated error bound and confidence level. We provide a versatile framework that allows users to perform localization queries based on the required accuracy and certainty. Ensuring coverage of a wireless sensor network is critical in many applications. Most schemes that analyze and implement coverage assume that sensor locations are known. For large sensor networks, errors occur when estimating node positions. Based on the errors that arise from the localization process, we present coverage algorithms that associate a confidence level with the coverage. We introduce a system that handles user coverage queries based on the coverage and certainty a situation requires.

With the emergence of wireless sensor networks and pervasive computing, innovative location-aware technologies and services are being investigated. Several iterative approaches employing distributed computations over sensors have been proposed in the literature for locating all the sensor nodes in the network. Due to their iterative nature these techniques are inefficient in terms of power, a very precious resource in sensor networks. This paper presents a novel power efficient approach aimed at identifying the locations of all the nodes in a sensor network given the location of a small subset of nodes. The technique, using evolution strategies, is independent of the ranging method used to estimate distances between nodes and involves sink nodes in the computation. The proposed approach provides substantial energy savings over existing techniques while providing comparable accuracy, and requires the presence of at least one neighbor for each sensor node compared to at least 3 neighbors for most of the existing techniques

Deployment of a vast array of tiny smart sensors (sensor devices with processors) interconnected over wireless channels are enabling their pervasive use in a variety of defense and commercial applications, such as environmental monitoring (e.g. traffic, habitat, security), industrial sensing and diagnostics (e.g. factory, appliances), infrastructures (e.g. power grid, water distributions, waste disposal), and battlefield awareness (e.g. multi-target tracking). While the task of developing and implementing pervasive applications for such a scenario is exciting, it poses tremendous challenges. In this context, design of algorithms for processing information in sensor networks is an emerging research area. However, due to the unique characteristics of the sensor network computing paradigm, the objective is to design algorithms that meet the two contradicting goals of low power and fast executions times, i.e., the algorithms should be energy efficient and they should exploit the parallel computing resources available in sensor nodes. In this paper, we present a power-time efficient algorithm for computing Fast Fourier Transform over data distributed across smart sensors. The Fast Fourier Transform (FFT) [2] has been studied extensively as a frequency analysis tool in diverse application areas such as audio, signal, and image processing, and several other real time data applications [5].

MEMS technology has improved such that the capabilities of large sensor devices can now be encompassed in devices that are the size of a penny. These resource constraint devices are confronted with the challenges of ensuring accuracy of observations while conserving power resources. This paper analyzes the two modes of collaborative signal processing employed in sensor networks, namely, data fusion and decision fusion, to achieve greater accuracy. A breakdown of the computation and communication complexity is presented for each fundamental step of data and decision fusion processes, to help in identifying features for possible optimization. We show that a tradeoff exists between the two approaches depending on the underlying application domain.

Locating objects, such as keys, office equipment, people, or even an enemy vehicle, is an application that has received a lot of attention over recent years. With the emergence of wireless networks and mobile computing devices, providing location-aware technology and services to new applications has become important for developers. Recent advances in sensor technology have allowed wireless sensor networks to provide location services. Many applications of wireless sensor networks assume that the devices are location-aware. In this paper, we discuss the Ferret localization system. Using only the radio features of the sensors, the Ferret system provides two techniques for locating an object. The system provides good results, but several extensions are discussed to make it more scalable and reliable.