Friday, February 24, 2012
Reminder: paper submission deadline for UCNC 2012
A reminder that the deadline for submitting papers to the 11th Conference on Unconventional Computation and Natural Computation (UCNC) 2012 is 26 March 2012. This conference will be held in Orleans, France, 3-6 September, 2012.
Thursday, February 23, 2012
Reminder: paper submission deadline for CIDM 2012
A reminder that the deadline for submitting papers to the Third International Workshop on Computational Intelligence for Disaster Management (CIDM) 2012 is March 25 2012. This workshop will be held in Bucharest, Romania, 19-21 September 2012.
Wednesday, February 22, 2012
Using MLP to model the distribution of bacterial crop diseases
A new paper I co-authored with Sue Worner at Lincoln University is now available and describes how we used MLP to model the global distribution of bacterial crop diseases.
We had data on the presence or absence of certain species of bacterial crop disease (that is, bacteria that infect and cause diseases in plants we use as crops) in 459 geo-political regions throughout the world. We also had data on the climate in these regions and the presence of host plant species. We created MLP that predicted the presence or absence of the bacteria species from climate (abiotic factors). We also created MLP that predicted the presence or absence of the bacteria species from the host plant species assemblages (biotic factors). While both of these approaches worked, we got much better accuracies by combining the outputs into ensembles, and by using a cascaded or tandem ANN approach.
Ensembles are a way of combining the outputs of several ANN. An input vector is propagated through each of the ANN, and the output values combined either statistically (the final output value is the max, mean or median of the uncombined outputs) or algorithmically (output is determined as a majority vote of the uncombined outputs - that is, if the majority of the values is above a threshold, the output of the ensemble is a presence, otherwise, absence). We looked at three different kinds of ensemble: firstly, ensembles of the best ten MLP trained on abiotic inputs; secondly, ensembles of the best ten MLP trained on biotic inputs; and finally, ensembles that combined the best ten MLP trained on abiotic input as well as the best ten MLP trained on biotic inputs. These last ensembles were particularly interesting, as it allowed us to make predictions of species distributions using both biotic and abiotic factors simultaneously. The rationale behind ensembles is that different MLP learn different parts of the problem space: by combining the outputs of several MLP, it is possible to cover a larger part of the problem space, and therefore to boost prediction accuracy. Combining abiotic and biotic factors is the same idea. We know that an organism is affected by both of these factors, so combining both of them allows us to make more accurate predictions.
While the ensemble approach boosted the prediction accuracies, we thought we could do better, so we created MLP that took as inputs the outputs of the very best MLP trained on abiotic and biotic factors. In other words, the outputs of the climate and host networks were used as the input values for a second-level of MLP, which were then trained on the presence and absence of the bacteria species. The idea behind tandem ANN is that, if a first-level network makes a mistake - that is, if a climate or host MLP makes an incorrect prediction - then the tandem network can learn to correct it. Again, we were combining abiotic and biotic factors to make predictions.
The results of all these techniques were that while the single-level MLP were able to predict the distributions of the crop diseases fairly well, combining abiotic and biotic factors gave much better accuracy, whether the combination was achieved by a simple ensemble approach, or by using a tandem MLP approach.
This paper is published in Computational Ecology and Software, an open-access journal. Given my previous posts extolling the virtues of open access journals (see here, here, here and here) I'm putting my academic money where my mouth is, and submitting to open-access journals.
We had data on the presence or absence of certain species of bacterial crop disease (that is, bacteria that infect and cause diseases in plants we use as crops) in 459 geo-political regions throughout the world. We also had data on the climate in these regions and the presence of host plant species. We created MLP that predicted the presence or absence of the bacteria species from climate (abiotic factors). We also created MLP that predicted the presence or absence of the bacteria species from the host plant species assemblages (biotic factors). While both of these approaches worked, we got much better accuracies by combining the outputs into ensembles, and by using a cascaded or tandem ANN approach.
Ensembles are a way of combining the outputs of several ANN. An input vector is propagated through each of the ANN, and the output values combined either statistically (the final output value is the max, mean or median of the uncombined outputs) or algorithmically (output is determined as a majority vote of the uncombined outputs - that is, if the majority of the values is above a threshold, the output of the ensemble is a presence, otherwise, absence). We looked at three different kinds of ensemble: firstly, ensembles of the best ten MLP trained on abiotic inputs; secondly, ensembles of the best ten MLP trained on biotic inputs; and finally, ensembles that combined the best ten MLP trained on abiotic input as well as the best ten MLP trained on biotic inputs. These last ensembles were particularly interesting, as it allowed us to make predictions of species distributions using both biotic and abiotic factors simultaneously. The rationale behind ensembles is that different MLP learn different parts of the problem space: by combining the outputs of several MLP, it is possible to cover a larger part of the problem space, and therefore to boost prediction accuracy. Combining abiotic and biotic factors is the same idea. We know that an organism is affected by both of these factors, so combining both of them allows us to make more accurate predictions.
While the ensemble approach boosted the prediction accuracies, we thought we could do better, so we created MLP that took as inputs the outputs of the very best MLP trained on abiotic and biotic factors. In other words, the outputs of the climate and host networks were used as the input values for a second-level of MLP, which were then trained on the presence and absence of the bacteria species. The idea behind tandem ANN is that, if a first-level network makes a mistake - that is, if a climate or host MLP makes an incorrect prediction - then the tandem network can learn to correct it. Again, we were combining abiotic and biotic factors to make predictions.
The results of all these techniques were that while the single-level MLP were able to predict the distributions of the crop diseases fairly well, combining abiotic and biotic factors gave much better accuracy, whether the combination was achieved by a simple ensemble approach, or by using a tandem MLP approach.
This paper is published in Computational Ecology and Software, an open-access journal. Given my previous posts extolling the virtues of open access journals (see here, here, here and here) I'm putting my academic money where my mouth is, and submitting to open-access journals.
Tuesday, February 21, 2012
Conference paper deadline: NIPS 2012
The deadline for submitting papers to Neural Information Processing Systems (NIPS) 2012 is 2 June 2012. This conference will be held at Lake Tahoe, Nevada, 3-6 December, 2012.
Monday, February 20, 2012
Call for papers: Applications of ECoS
Special issue of Evolving Systems on
Applications of Evolving Connectionist Systems
Guest Editor
Scope
The topic of this special issue is “Applications of Kasabov’s Evolving Connectionist Systems”.
In modern society, the volume and rate of data production are huge and set to increase. To process and utilise this avalanche of data, methods are needed that can rapidly and accurately model it as it becomes available. These models must be able to learn throughout their lifetimes, without forgetting what they have previously learned, and be able to explain themselves.
Kasabov’s Evolving Connectionist Systems (ECoS) are able to fulfil each of these requirements. They are a class of constructive neural networks that learn via structural growth and adaptation. They have a fast, one-pass learning algorithm, where all that can be learnt from the data is learned in the first training pass. Because of their open structure, they exhibit continuous, life-long learning whereby the structure expands as necessary to accommodate new data. Finally, they have a strong resistance to catastrophic forgetting following additional training on new data.
Examples of ECoS networks include the Evolving Fuzzy Neural Network (EFuNN), which was the first ECoS network published and is characterised by embedded fuzzy logic elements. There is also the Simple Evolving Connectionist System (SECoS), which is essentially an EFuNN with the fuzzy elements removed, and the Dynamic Evolving Fuzzy Inference System (DENFIS) for discovering Takagi-Sugeno style fuzzy rules. Many ECoS networks use fuzzy rule extraction algorithms that allow for the explanation of what the networks have learned, in a comprehensible manner.
ECoS networks are well suited to applications that are dealing with new data continuously and that have dynamic, time-critical aspects. Previous applications of ECoS include:
- Stock market prediction and macroeconomic modelling
- Speech recognition, especially multi-speaker speech recognition
- Bioinformatics and medical modelling
- Image and video parsing
- Robot control
- Information system security
The special issue is concerned with all aspects of the application of ECoS networks to real-life problems and data sets. Topics of interest include, but are not limited to:
- Applications of ECoS to real-world problems
- Data mining of complex data sets using ECoS
- Comparisons of ECoS with other algorithms over real-world data sets
- Modifications of ECoS algorithms to fit them to real-world problems
Proposed Schedule
- Submission due date: 16 April, 2012
- Preliminary notification of acceptance: 4 June, 2012
- Revised manuscripts due: 9 July, 2012
- Final acceptance notification: 6 August, 2012
- Final version due: 3 September, 2012
- Intended publication date: January, 2013
Submission
The special issue invites original contributions within the specified scope. Manuscripts must not be under review elsewhere, nor can they have been previously published. Extended conference papers must contain at least 30% new material. Please format all manuscripts according to the Instructions for Authors:
Please submit all papers via the online submission system:
Friday, February 17, 2012
Reminder: paper submission deadline for IEEE CISDA 2012
A reminder that the deadline for papers submitted to the IEEE Workshop on Computational Intelligence for Security and Defence Applications (IEEE CISDA) 2012 is 19 March 2012. This workshop will be held in Ottawa, Canada, 11-13 July 2012.
Thursday, February 16, 2012
Call for papers: CIDU 2012
The deadline for submitting papers to the 2012 Conference on Intelligent Data Understanding (CIDU) 2012 is 18 May 2012. This conference will be held in Boulder, Colorado, 24-26 October, 2012.
Wednesday, February 15, 2012
Reminder: paper submission deadline ICONIP 2012
A reminder that the paper submission deadline for the International Conference on Neural Information Processing (ICONIP) 2012 is May 15, 2012. This conference will be held in Doha, Qatar, November 26-29, 2012.
Tuesday, February 14, 2012
Reminder: paper submission deadline for IEEE CIMSA 2012
A reminder that the deadline for submitting papers to the IEEE International Conference on Computational Intelligence for Measurement Systems and Applications (IEEE CIMSA) 2012 is 15 March 2012. This conference will be held in Tianjin, China, 2-4 July 2012.
Monday, February 13, 2012
Reminder: paper submission deadline for PPSN 2012
A reminder that the deadline for submitting papers to the 12th International Conference on Parallel Problem Solving from Nature (PPSN) 2012 is the 12th of March, 2012. This conference will be held in Taormina, Italy, 1-5 September 2012.
Friday, February 10, 2012
Paper deadline extension: ICIC 2012
The deadline for papers submitted to the International Conference on Intelligent Computing (ICIC) 2012 has been extended to March 31, 2012. This conference will be held in Huangshan, China, July 25-29, 2012.
Conference paper submission deadline for UKCI 2012
The paper submission deadline for the 12th UK Annual Workshop on Computational Intelligence (UKCI) 2012 is May 1, 2012. This workshop will be held in Edinburgh, Scotland, UK, 5-7 September, 2012.
Thursday, February 9, 2012
New article: IEEE CIS social media
An article I co-wrote with Albert Lam, Dongrui Wu and Pablo Estevez, describing the social media presence of the IEEE Computational Intelligence Society:
Via IEEE Xplore.
Alternative link.
Via IEEE Xplore.
Alternative link.
Conference paper deadline: iCAST 2012
The deadline for submitting papers to the 4th International Conference on Awareness Science and Technology (iCAST) 2012 is 15 April 2012. The conference will be held in Seoul, Korea, 21-24 August, 2012.
Wednesday, February 8, 2012
Call for papers: AI'12
The deadline for submitting papers to the 25th Australasian Joint Conference on Artificial Intelligence (AI) 2012 is 29 June, 2012. This conference will be held in Sydney, Australia, 4-7 December, 2012.
Tuesday, February 7, 2012
Reminder: Paper submission deadline for ECAI 2012
A reminder that the deadline for papers submitted to the 20th European Conference on Artificial Intelligence (ECAI) 2012 is 6 March 2012. This conference will be held in Montpellier, France, 27-21 August, 2012.
Subscribe to:
Posts (Atom)
Posts
