Alessio Gaspar

Two models of immunization for time dependent optimization

This paper investigates the relevance of the immune system metaphor to time-dependent optimization (TDO). First, we review previous results underlining the over-average reactiveness and robustness featured by Sais (Simple Artificial Immune System) in comparison to well-known evolutionist approaches. Then, we evaluate its immunization capability (i.e. improving its robustness to previously encountered optima) and eventually propose Yasais (Yet Another Simple Artificial Immune System).

An Evolutionary Approach for Time Dependent Optimization

Many real-world problems involve measures of objectives that may be dynamically optimized. The application of evolutionary algorithms, such as genetic algorithms, in time dependent optimization is currently receiving growing interest as potential applications are numerous ranging from mobile robotics to real time process command. Moreover, constant evaluation functions skew results relative to natural evolution so that it has become a promising gap to combine effectiveness and diversity in a genetic algorithm. This paper features both theoretical and empirical analysis of the behavior of genetic algorithms in such an environment. We present a comparison between the effectivenss of traditional genetic algorithm and the dual genetic algorithm which has revealed to be a particularly adaptive tool for optimizing a lot of diversified classes of functions. This comparison has been performed on a model of dynamical environments which characteristics are analyzed in order to establish the basis of a testbed for further experiments. We also discuss fundamental properties that explain the effectiveness of the dual paradigm to manage dynamical environments.

Time dependent optimization with a folding genetic algorithm

Time-dependent optimization has revealed to be a promising gap for the entire genetic algorithms community since it has numerous applications. This paper extends previous work (Collard et al., 1996) related to the use of meta-genes in the so-called dual genetic algorithms (DGAs). A more generic framework, involving a variable number of genes, is introduced. Folding genetic algorithms are thus proposed as a new class of genetic algorithms, whose effectiveness is investigated on two well-known models of dynamical environments and compared to simple genetic algorithms and DGAs. Eventually, further analysis of these results enlightens the ability of FGAs to evolve a metric over the search space (i.e. a kind of encoding scheme) along with potential solutions. These particularly encouraging results open up interesting perspectives, as FGAs could be applied to to other fundamental problems investigated by the genetic algorithms community in order to measure the benefits of this really meta-level of evolution.

Rapid conversion of an IT degree program to online delivery: impact, problems, solutions and challenges

At our institution, online delivery was initially seen as a way to extend the geographic reach of a specialized upper-division degree program, the Bachelor of Science in Applied Sciences (BSAS), targeting holders of community college Associates in Science (AS) degrees, with the hoped-for impact of increasing enrollment. This program shared many required courses with our flagship program, the BS in Information Technology (BSIT), including the upper-division entry point for both programs, CGS3303 IT Concepts. Implementing this course in an asynchronous online mode resulted in an increase in enrollment from 17 (fall, 2007) to 70 (spring, 2008); most of these 70 students identified as BSIT, leading us to fast-track conversion to online modes for all of our core courses and electives. This accelerated pace of online development presented many challenges and difficulties, including ensuring the quality of our finished course product. Faculty buy-in, assisted by a stipend program for online development, was essential, as was the presence of quality instructional technology support. An NSF-funded infrastructure (SOFTICE) allowed the offering of problematic laboratory courses (Operating Systems, Networks) safely and securely in an online mode. A course quality review process was implemented as part of the stipend program, which also largely resolved intellectual property issues. Nonetheless, remaining are some issues and concerns, which are discussed in this paper. Recommendations, observations and suggested processes are included for those who may be considering transitioning an IT degree program to online delivery.

There is ALife beyond convergence: using a dual sharing to adapt in time dependent optimization

Using a time dependent optimization problem as a benchmark we propose to improve the genetic algorithms adaptiveness by an appropriate diversity management dynamics. The so-called Dual Sharing GA is an extension of a former algorithm based on the use of a genotype to phenotype mapping. We experimentally evaluate the benefits over similar approaches to diversity management. We also analyze more precisely the dynamics exhibited at both phenotypic and genotypic levels in comparison with our previous approach.

SOFTICE undergraduate operating systems laboratories

1 This material is based in part upon work supported by the National Science Foundation under award number 0410696. Any opinions, findings, and conclusions or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. The hands-on experience provided by an OS laboratory can significantly improve real operating systems’ understanding. Despite their inherent complexity, such systems can clarify lectures by exposing the implementation constraints at the origin of the lectured concepts, algorithms and tradeoffs. Such a pedagogical approach helps students to “make sense” of facts instead of merely memorizing them thus resulting in a more active learning dynamics. As of today, the availability of the Linux kernel source code, its increasing presence in server-side industries and its many related publications and pedagogical material have made it a tool of choice in achieving such an endeavor. As always, there is still room for improvement.

Differences in the Learning Principles Dominating Student-Student vs. Student-Instructor Interactions while Working on Programming Tasks

Peer learning principles have been successfully applied to novice programmers. Pedagogies such as Pair Programming, Peer Testing, Peer review of code or tests, or, more generally Peer Instruction, have repeatedly demonstrated their effectiveness in improving both individual performance and retention rates. This paper proposes to supplement the existing literature by investigating how students interact with one another during collaborative programming tasks. More specifically, we are interested in comparing the learning principles used during student-student interactions with those used during student-instructor or student-teaching assistant dialogs. Students in online and face to face courses, who worked collaboratively on programming assignments, were surveyed to gain an understanding of the frequency with which they engaged in specific activities. These that are representative of the learning principles that have been supported by research to promote learning. Results suggest that some learning principles, may be absent from student-student interactions. We discuss how the success of collaborative programming pedagogies put into question the role of these principles and whether they may contribute to further improve peer-based approaches.

Multi-perspective survey of the relevance of the revised bloom's taxonomy to an introduction to linux course

Equipping students with higher-order thinking skills as part of a program in information technology is no trivial aim. Course creation must always have this goal in mind. In particular, learning activities and assessments must be designed to teach, encourage the use of, and assess success in achieving this goal beyond merely teaching facts, methods and techniques. In this paper, we examine the degree to which we were able to assess higher-order thinking skills in students enrolled in the first course of an online Linux system administration curriculum. To assist other educators contemplating similar efforts, we briefly describe methods used to classify quiz and assignment items using the Revised Bloom Taxonomy (RBT) and discuss results from a survey administered to students who completed the course. Lessons learned throughout the process are described.

Evolutionary Practice Problems Generation: Design Guidelines

This paper identifies design guidelines for the application of evolutionary techniques to the task of generating practice problems for learners in an Intelligent Tutoring System. To this end, we designed experiments that progressively incorporated an increasing number of the characteristics we expect to find in our target application. These features included noisy evaluations, overspecialization, and the need to mitigate user fatigue resulting from interactive evaluations of practice problems. As we did so, we evaluated the potential of recent breakthroughs in coevolutionary learning theory and identified the tradeoff specific to educational applications.