Andrea Pohoreckyj Danyluk

Finding new rules for incomplete theories: explicit biases for induction with contextual information

Publisher Summary This chapter reviews two disparate machine learning approaches that have seen rigorous review: (1) explanation-based learning (EBL) and (2) similarity-based learning (SBL). EBL is a deductive approach in which a definition of a concept is learned usually after observing only a single example of that concept. When EBL cannot derive a complete explanation, the partial explanation forms a context in which learning takes place. EBL assumes that a system is given an explicit theory of the domain that is complete, correct, and tractable. SBL is an empirical technique that involves the comparison of a large number of input examples. SBL suffers because of its lack of an explicit domain theory. Information extracted from partial explanations and from complete explanations, can be exploited by SBL to do better induction of the missing domain knowledge. The extracted information constitutes a strong bias for SBL. EBL and SBL have been applied to problems in a variety of domains. Both methods are rife with well-defined problems.

Event-driven programming is simple enough for CS1

We have recently designed a CS 1 course that integrates event-driven programming from the very start. Our experience teaching this course runs counter to the prevailing sense that these techniques would add complexity to the content of CS 1. Instead, we found that they were simple to present and that they also simplified the presentation of other material in the course. In this paper, we explain the approach we used to introduce event-driven methods and discuss the factors underlying our success.

Introducing concurrency in CS 1

Because of the growing importance of concurrent programming, many people are trying to figure out where in the curriculum to introduce students to concurrency. In this paper we discuss the use of concurrency in an introductory computer science course. This course, which has been taught for ten years, introduces concurrency in the context of event-driven programming. It also makes use of graphics and animations with the support of a library that reduces the syntactic overhead of using these constructs. Students learn to use separate threads in a way that enables them to write programs that match their intuitions of the world. While the separate threads do interact, programs are selected so that race conditions are generally not an issue.

Why structural recursion should be taught before arrays in CS 1

The approach to teaching recursion in introductory programming courses has changed little during the transition from procedural to object-oriented languages. It is still common to present recursion late in the course and to focus on traditional, procedural examples such as calculating factorials or solving the Towers of Hanoi puzzle. In this paper, we propose that the shift to object-oriented programming techniques calls for a significant shift in our approach to teaching recursion. First, we argue that in the context of object-oriented programming students should be introduced to examples of simple recursive structures such as linked lists and methods that process them, before being introduced to traditional procedural examples. Second, we believe that this material should be presented before students are introduced to structures such as arrays. In our experience, the early presentation of recursive structures provides the opportunity to reinforce the fundamentals of defining and using classes and better prepares students to appreciate the reasons to use classes to encapsulate access to other data structures when they are presented.

Feature Selection vs Theory Reformulation: A Study of Genetic Refinement of Knowledge-based Neural Networks

Expert classification systems have proven themselves effective decision makers for many types of problems. However, the accuracy of such systems is often highly dependent upon the accuracy of a human experts domain theory. When human experts learn or create a set of rules, they are subject to a number of hindrances. Most significantly experts are, to a greater or lesser extent, restricted by the tradition of scholarship which has preceded them and by an inability to examine large amounts of data in a rigorous fashion without the effects of boredom or frustration. As a result, human theories are often erroneous or incomplete. To escape this dependency, machine learning systems have been developed to automatically refine and correct an experts domain theory. When theory revision systems are applied to expert theories, they often concentrate on the reformulation of the knowledge provided rather than on the reformulation or selection of input features. The general assumption seems to be that the expert has already selected the set of features that will be most useful for the given task. That set may, however, be suboptimal. This paper studies theory refinement and the relative benefits of applying feature selection versus more extensive theory reformulation.

Event-driven programming facilitates learning standard programming concepts

We have designed a CS 1 course that integrates event-driven programming from the very start. In cite BDMITiCSE1 we argued that event-driven programming is simple enough for CS 1 when introduced with the aid of a library that we have developed. In this paper we argue that early use of event-driven programming makes many of the standard topics of CS 1 much easier for students to learn by breaking them into smaller, more understandable concepts.

Experiences mapping and revising curricula with CS2013

1. SUMMARY Roughly once per decade, the ACM and IEEE-Computer Society form a joint task force to survey the discipline and produce curricular guidelines for undergraduate computer science programs. The latest guidelines document, Computer Science Curricula 2013 (CS2013), was released in the Fall of 2013 after multiple rounds of public review and feedback. A significant novel feature of CS2013 is the structuring of the Body of Knowledge core topics into two tiers. While Core-Tier1 topics are expected to be included in every undergraduate computer science program, individual programs may choose to cover only a subset of Core-Tier2 topics, depending upon institutional size, resources, and goals. CS2013 recommends at least 80% coverage of CoreTier2, which provides flexibility for designing curricula that meet institutional constraints while still maintaining disciplinary standards.

ACM/IEEE computer science 2013 exemplar-fest

Beginning with the publication of Curriculum 68, ACM and IEEE-Computer Society have sponsored various efforts to establish international curricular guidelines for undergraduate programs in computing. Work on the next volume, Computer Science 2013 is well underway, with the Ironman draft out shortly before SIGCSE 2013. The Ironman draft includes course and curricular exemplars, which should serve as a rich resource for those trying to meet the curriculum standards. This special session highlights the exemplar section of the Ironman report through a description of its purpose, presentation of several exemplars, and an invitation to the SIGCSE community to participate by submitting exemplars and providing feedback on what they would find useful in this section of the CS 2013 final report.

SIGCSE 2018 new educator workshop

A successful career as a computer science educator involves more than a deep understanding of a research area. Yet many new CS educators experience relatively little educator training - and face more questions than answers, e.g., What different career-path choices do CS educators pursue? How do I choose an institution or career path that is right for me? How can I balance teaching, research, service, and a life beyond all those things? What are the balancing acts involved in working effectively with colleagues and managing the advancement and tenure process? What tips could help me organize a course, deliver engaging lectures, and build lasting relationships with students? The New Educators Workshop (NEW) is a pre-symposium event at SIGCSE 2018 that, through presentations, discussions, and small-group community building, will tackle all of these questions. NEW is designed to assist aspiring and early-career educators in exploring the non-research facets of an academic career. It will run on February 21, 2018 from 9am to 5pm, and is open to graduate students who are considering teaching-related careers, as well as pre-tenure faculty members seeking guidance and/or networking support. The 2018 NEW is one of several career-focused, SIGCSE-affiliated workshops that have served more than 200 educators over the past decade.

Predicting Expressive Bow Controls for Violin and Viola

Though computational systems can simulate notes on a staff of sheet music, capturing the artistic liberties professional musicians take to communicate their interpretation of those notes is a much more difficult task. In this paper, we demonstrate that machine learning methods can be used to learn models of expressivity, focusing on bow articulation for violin and viola. First we describe a new data set of annotated sheet music with information about specific aspects of bow control. We then present experiments for building and testing predictive models for these bow controls, as well as analysis that includes both general metrics and manual examination.