please note that the final order of presentations may change
Thursday 12 April
Technology Provider Presentations (09:30–11:00)
This session will be devoted to presenting solutions that the evolutionary computation community has to offer to Industry. Each technology provider will be given 15 minutes to present the main idea of their solution.
Evolutionary Sound Synthesis of Birdsongs
José Fornari This talk presents a computer model that creates a soundscape of realistic birdsongs. The sound is never repeated but always keep a similarity. The model could be used, for example, in large indoor environments (such as shopping malls) to create an artificial (yet realistic) sonic experience resembling nature. A sample of this artificial soundscape can be heard at: http://www.4shared.com/audio/gEsDwkNw/soundsample.html.
File Type Detection using GP
Ahmed Kattan The author has used GP to analyze the binary streams in the files and detect their datatypes (e.g., txt, pdf, gif, …etc.). This algorithm can have many useful applications. For example, it can be used to check emails’ attachments of possible malicious files. Also, this can be used to enforce LAN policies (e.g., a company don’t allow sending PDFs out of its LAN). In a nutshell, any real-world application that requires the user to analyse many files and distinguish their contexts, my algorithm can be used to automate this process.
Cloud Based Machine Learning for Biotechnology
Paul Walsh Researchers in the Cork Institute of Technology have developed a cloud based management system for a range of biotechnology problems from gene identification to protein structure prediction. The system provides a suite of machine learning algorithms implemented in the cloud to provide researchers in academic and industry alike to perform research more efficiently. Case studies show that this system can reduce the time needed to conduct research by up to 80%.
Voluntary Optimization via Mobile Devices
Antonio J. Fernández Leiva This talk presents a concept that is associated to a patent of us. In particular we want to show a model for “voluntary optimization” via mobile devices (i.e., handheld). The model might be of interest to any telecommunication company.
Evolutionary Verification of Hardware and Software Devices
Alberto Tonda Human capability for designing complex systems outperforms our ability to attest their correct behavior. One of the most striking examples is microprocessor production, where very few chips actually work as expected in the first release; but the same is true for modern user-end operative systems, where multiple applications designed by different teams work in the same environment, with unpredictable interactions. Thanks to our experience in real-world industrial problems, developed during projects in cooperation with companies such as Intel, Motorola and STMicroelectronics, we propose an evolutionary technique able to attest the correct behavior of complex systems, with the aim to integrate an existing qualifying verification plan. The methodology is completely automatic, so it can be run in parallel with the action of verification engineers, for maximum performance; moreover, due to the nature of evolutionary computation, it will explore patterns that humans are unlikely to intentionally provide, but that could appear due to accidents or errors; and, finally, since we rely upon a general-purpose state-of-the-art genetic programming approach, little to no additional coding will be needed to adjust the methodology to different applications.
Kimeme: A new flexible platform for multi-objective and multi-disciplinary optimization
Ernesto Mininno In all fields of expertise, optimization is essential. Many practical applications, e.g. in engineering, chemistry, logistic, or finance, often require the solution of an optimization problem. To fill the knowledge gap between domain experts and optimization scientists, we developed Kimeme, a new flexible platform for multi-disciplinary optimization. Kimeme is a unique platform that can be used both for algorithm and process design. It combines an advanced graphical environment, a comprehensive set of post-processing tools, and an open-source library of state-of-the-art single and multi-objective optimization algorithms. In a memetic fashion, algorithms are decomposed into operators, so that users can easily create new optimization methods, just combining built-in operators or creating new ones with Java and Python. Similarly, the optimization process is described according to a data-flow logic, so that it can be seamlessly integrated with external software, such as CAD/CAE packages, Matlab, spreadsheet programs, scripts, and custom pieces of object-oriented code. In addition to that, Kimeme includes a native distributed computing framework, particularly suited for computationally expensive processes, which allows parallel computations at different level of granularity (from a single solution to a complete algorithm run) on clusters and heterogeneous LANs. Case studies from our industrial and academic partners show that Kimeme is remarkably faster than similar commercial and free platforms, still being able to find comparable or better solutions on highly complex optimization problems.
Thursday 12 April
Technology User Presentations, Demos, and Discussion
(11:30–13:00)
The aim of this session is to get technology providers and users to speak to each other. This aim will be pursued as follows:
Industrial participants wishing to do so will be given the stand to briefly illustrate a problem they face and for which they are seeking a solution.
Demos of some of the solutions presented in Session 1 will be presented.
Face-to-face meetings and discussion will take place among participants; these will continue informally after the end of the session, if necessary.
draft programme evotransfer 2012
please note that the final order of presentations may change
Thursday 12 April
Technology Provider Presentations (09:30–11:00)
This session will be devoted to presenting solutions that the evolutionary computation community has to offer to Industry. Each technology provider will be given 15 minutes to present the main idea of their solution.
Evolutionary Sound Synthesis of Birdsongs
José Fornari
This talk presents a computer model that creates a soundscape of realistic birdsongs. The sound is never repeated but always keep a similarity. The model could be used, for example, in large indoor environments (such as shopping malls) to create an artificial (yet realistic) sonic experience resembling nature. A sample of this artificial soundscape can be heard at: http://www.4shared.com/audio/gEsDwkNw/soundsample.html.
File Type Detection using GP
Ahmed Kattan
The author has used GP to analyze the binary streams in the files and detect their datatypes (e.g., txt, pdf, gif, …etc.). This algorithm can have many useful applications. For example, it can be used to check emails’ attachments of possible malicious files. Also, this can be used to enforce LAN policies (e.g., a company don’t allow sending PDFs out of its LAN). In a nutshell, any real-world application that requires the user to analyse many files and distinguish their contexts, my algorithm can be used to automate this process.
Cloud Based Machine Learning for Biotechnology
Paul Walsh
Researchers in the Cork Institute of Technology have developed a cloud based management system for a range of biotechnology problems from gene identification to protein structure prediction. The system provides a suite of machine learning algorithms implemented in the cloud to provide researchers in academic and industry alike to perform research more efficiently. Case studies show that this system can reduce the time needed to conduct research by up to 80%.
Voluntary Optimization via Mobile Devices
Antonio J. Fernández Leiva
This talk presents a concept that is associated to a patent of us. In particular we want to show a model for “voluntary optimization” via mobile devices (i.e., handheld). The model might be of interest to any telecommunication company.
Evolutionary Verification of Hardware and Software Devices
Alberto Tonda
Human capability for designing complex systems outperforms our ability to attest their correct behavior. One of the most striking examples is microprocessor production, where very few chips actually work as expected in the first release; but the same is true for modern user-end operative systems, where multiple applications designed by different teams work in the same environment, with unpredictable interactions. Thanks to our experience in real-world industrial problems, developed during projects in cooperation with companies such as Intel, Motorola and STMicroelectronics, we propose an evolutionary technique able to attest the correct behavior of complex systems, with the aim to integrate an existing qualifying verification plan. The methodology is completely automatic, so it can be run in parallel with the action of verification engineers, for maximum performance; moreover, due to the nature of evolutionary computation, it will explore patterns that humans are unlikely to intentionally provide, but that could appear due to accidents or errors; and, finally, since we rely upon a general-purpose state-of-the-art genetic programming approach, little to no additional coding will be needed to adjust the methodology to different applications.
Kimeme: A new flexible platform for multi-objective and multi-disciplinary optimization
Ernesto Mininno
In all fields of expertise, optimization is essential. Many practical applications, e.g. in engineering, chemistry, logistic, or finance, often require the solution of an optimization problem. To fill the knowledge gap between domain experts and optimization scientists, we developed Kimeme, a new flexible platform for multi-disciplinary optimization. Kimeme is a unique platform that can be used both for algorithm and process design. It combines an advanced graphical environment, a comprehensive set of post-processing tools, and an open-source library of state-of-the-art single and multi-objective optimization algorithms. In a memetic fashion, algorithms are decomposed into operators, so that users can easily create new optimization methods, just combining built-in operators or creating new ones with Java and Python. Similarly, the optimization process is described according to a data-flow logic, so that it can be seamlessly integrated with external software, such as CAD/CAE packages, Matlab, spreadsheet programs, scripts, and custom pieces of object-oriented code. In addition to that, Kimeme includes a native distributed computing framework, particularly suited for computationally expensive processes, which allows parallel computations at different level of granularity (from a single solution to a complete algorithm run) on clusters and heterogeneous LANs. Case studies from our industrial and academic partners show that Kimeme is remarkably faster than similar commercial and free platforms, still being able to find comparable or better solutions on highly complex optimization problems.
Thursday 12 April
Technology User Presentations, Demos, and Discussion
(11:30–13:00)
The aim of this session is to get technology providers and users to speak to each other. This aim will be pursued as follows: