2008: Natural Computing: The Virtual Laboratory and Two Real-World Applications
Download as PPT, PDF0 likes909 views
The document summarizes natural computing and provides two examples of its real-world applications. It introduces natural computing as computing inspired by nature, simulating natural phenomena, and using natural materials for computing. It then describes a virtual laboratory on natural computing and its features. Finally, it discusses two applications: an automatic grain classifier using computer vision and pattern recognition, and a container terminal scheduling system using an artificial immune network algorithm to optimize crane movements.
![Natural Computing : The Virtual Laboratory and Two Real-World Applications Leandro Nunes de Castro [email_address] ; [email_address] Mackenzie University NatComp – From Nature to Business](https://image.slidesharecdn.com/conaiscorrect-120203073207-phpapp01/85/2008-Natural-Computing-The-Virtual-Laboratory-and-Two-Real-World-Applications-1-320.jpg)
2008: Natural Computing: The Virtual Laboratory and Two Real-World Applications
- 1. Natural Computing : The Virtual Laboratory and Two Real-World Applications Leandro Nunes de Castro [email_address] ; [email_address] Mackenzie University NatComp – From Nature to Business
- 2. Agenda Natural Computing Introduction and Motivation Computing Inspired by Nature Synthesis of Nature by Means of Computing Novel Computing Devices The Virtual Laboratory on Natural Computing Main Features and Where to Find Two Real World Applications A Grain Classifier Container Scheduling
- 3. Natural Computing Motivation and An Overview
- 4. Imagine a world where computers can create new universes, and within these universes there are natural forms that reproduce, grow and adapt. Imagine natural patterns, mountains, ant colonies, immune systems and brains, all learning and evolving, and becoming increasingly more adapted to the environment. Imagine if our computers could contain new forms of life. Think how this would affect our lives. Maybe we could automatically create house and music design, new forms of protecting computers against invaders, new forms of solving complex problems, new organisms and new forms of computing. Now stop imagining. Welcome to Computing in the New Millennium. Welcome to the Natural Computing age! Adapted from Digital Biology, by P. Bentley.
- 5. Current Computer Technology Turing Machines (TM) Computational device idealized by A. Turing in 1936 If a problem can be computed, then it can be computed by a Turing Machine J. von Neumman architecture
- 6. Main Features of Current Computers General-purpose machines Manipulate precisely precise information* Address-based memory Serial processing* Are not capable of generalizing Are not fault tolerant (robust) Are not adaptable* …
- 7. Questions Natural Computing help to Answer How to solve intractable problems? How to realistically synthesize natural phenomena? What computing devices will come next? N. of atoms per bit 2020: 1 atom per bit
- 8. Natural Computing: An Overview Nature x Computing Natural computing is the terminology used to encompass three paradigms: Computing inspired by nature The simulation and emulation of natural phenomena in computers Computing with natural materials
- 9. Natural Computing: An Overview
- 10. Computing Inspired by Nature Nature has evolved through ages in order to solve complex real-world problems Examples abound: nest building, nest cleaning, main senses (hearing, seeing, touching, smelling, tasting), etc. Computer algorithms based or inspired by nature have been developed for some time: Either to model nature, Or to solve complex real-world problems
- 11. Computing Inspired by Nature Neurocomputing Evolutionary Computing Swarm Intelligence Immunocomputing etc.
- 12. Synthesis of Nature by Means of Computing Biosciences: reductionist approach to understanding life Artificial Life & Fractal Geometry: bottom-up approach to synthesize life patterns and behaviors Focus on the computational synthesis of natural patterns and behaviors, not problem solving Widely used in computer graphics and movie making What is life? “ The property or quality that distinguishes living organisms from dead organisms and inanimate matter, manifested in functions such as metabolism, growth, reproduction, and response to stimuli or adaptation to the environment originating from within the organism. ” (Dictionary.com)
- 13. Synthesis of Nature by Means of Computing Artificial Life and Fractal Geometry
- 14. Novel Computing Devices If current computing technology will reach its limit in the near future, what would be the alternative material with which to compute? New computing methods based on other natural material than silicon: Molecules Membranes Quantum systems
- 15. Novel Computing Devices Quantum bit: | x = c 1|0 + c 2|1
- 16. Fundamentals of Natural Computing Some ideas that form the basis of natural computing: Capacity of dealing with complex problems The use of sets of candidate solutions Capacity of dealing imprecisely with imprecise information Robustness Distributivity Self-repair etc.
- 17. The Virtual Laboratory on Natural Computing LVCoN
- 18. Virtual Laboratories A virtual laboratory is a tool for distance learning and/or experimentation that allows people to share knowledge, data, voice, video, tools, and many other resources. It provides a suitable environment to extend, improve, integrate, refine, and assist the learning and/or experimentation process of many subjects, thus contributing to an increase of the effectiveness of scientific research and widening the use of scarce or costly equipments.
- 19. The Virtual Laboratory on Natural Computing Didactic contents Biological motivation, theoretical contents, pictures, references and pseudocodes. Simulations Java applets simulators available with a brief tutorial. Exercises with responses Sets of exercises with their respective answers to allow the students/instructors to evaluate themselves. http://lsin.unisantos.br/lvcon .
- 20. The Virtual Laboratory on Natural Computing
- 21. A Grain Classification Machine NatComp – From Nature to Business
- 22. Automatic Grain Classification Actors involved: Producers; Local and global consumers; Cooperatives; Banks; Stock Market. Motivation: Automatic certification of quality ; Avoid classification conflicts; No equivalent machine available***; Standardization.
- 23. Examples of Grain Defects
- 24. The Grain Classifier Project Public Investor, several commercial partners The Development Cycle: Conceptual Design Hardware Prototyping Computer Vision Pattern Recognition
- 25. Computer Vision Image Capture: Double face capture Feature Extraction: Color, Texture and Shape attributes Based on the RGB histograms Total of 70 attributes extracted per grain
- 26. Feature Selection and Classification Feature Selection: Filter and Wrapper Classification: Naïve Bayes KNN Support Vector Machines Multi-Layer Perceptrons aiNet+RBF SRABNET: Supervised RABNET
- 27. Experimental Results Estimating the Weight ICS-RBF = aiNet+RBF S tr : average deviation from the desired value (training) E tr :average estimation error for training (training)
- 28. Experimental Results Classification Performance E tr % E te % E CV % Std MLP 8,80 1,814 SVM 10,60 3,31 k-NN 15,10 3,00
- 29. Discussion The immune system approach demonstrated to be competitive Experiment with binary classification followed by defect classification Experiment hierarchical classification Possibility of automating the classification of grains Benchmark classification performance for humans: 95%
- 30. Operation Planning in a Container Terminal (CONTER) NatComp – From Nature to Business
- 31. The Importance of Container Terminals Most World commerce is performed using containers The operation of a CONTER is a very complicated and challenging task, involving space and equipment constraints, short time spans for ship docking, pre-specified ship plans, customs procedures, etc.
- 32. A Typical Problem: Scheduling RTGs When a Ship Plan is received in the terminal, the operators have to search and load the selected containers into the ship. The RTGs (Rubber Tyred Gantry Crane) are typical container handling equipments and move in three directions: x, y, and z. The less movements the RTGs make, the faster and cheaper becomes the ship loading.
- 33. RTGs Movements: Productive (a) (b) (c) (c) (d) (e)
- 34. RTGs Movements: Improductive (Set-Up) (a) (b) (c) (d)
- 35. Cost to Remove Containers where n x , n y and n z is the number of movements in direction X, Y and Z, respectively; V x , V y and V z is the RTG velocity in direction X, Y and Z, respectively; t T is the time spent to lock or unlock the spreader and n T is the number of spreader locking/unlocking.
- 37. A Demo on the RTG Scheduling Problem
- 38. RTG Scheduling Discussion Copt-aiNet was originally used for gene ordering Possibility of using other heuristics A few minutes are available to suggest the removal schedule This type of problem was rarely perceived as a ‘problem’ in container terminals Promotes over 40% reduction in set-up movements and time
- 39. Applied AIS: Discussion Vast number of applications available Great potential for further applications and developments Some issues that still deserve investigation: Formal aspects Comparison (theoretical and empirical) with other approaches Loads of testing: validation Real benefits How far to stretch the metaphor? Scalability Robustness to high dimensions
- 40. How Far Can We Go? THANK YOU FOR THE ATTENTION! QUESTIONS?