Conference
Computer Recognition Systems
About: Computer Recognition Systems is an academic conference. The conference publishes majorly in the area(s): Support vector machine & Image segmentation. Over the lifetime, 618 publications have been published by the conference receiving 3471 citations.
Papers
1 Jan 2005
TL;DR: Two approaches to deal with bias field corruption are discussed and the fuzzy c-means algorithm is modified so that it can be used to segment an MRI image corrupted by a bias field signal.
Abstract: Bias field signal is a low-frequency and very smooth signal that corrupts MRI images specially those produced by old MRI (Magnetic Resonance Imaging) machines. Image processing algorithms such as segmentation, texture analysis or classification that use the graylevel values of image pixels will not produce satisfactory results. A pre-processing step is needed to correct for the bias field signal before submitting corrupted MRI images to such algorithms or the algorithms should be modified. In this report we discuss two approaches to deal with bias field corruption. The first approach can be used as a preprocessing step where the corrupted MRI image is restored by dividing it by an estimated bias field signal using a surface fitting approach. The second approach shows how to modify the fuzzy c-means algorithm so that it can be used to segment an MRI image corrupted by a bias field signal.
114 citations
Proceedings Article•
23 May 2016TL;DR: In this article, the authors formulate the vDPI placement problem as a cost minimization problem and propose a centrality-based greedy algorithm to find the appropriate placement that meets the traffic management and operational cost constraints (license fees, network efficiency or power consumption).
Abstract: Network Functions Virtualization (NFV) is transforming how networks are operated and network services delivered. The network is more flexible and adaptable. In particular, virtual Deep Packet Inspection (vDPI) engines can be dynamically deployed as software on commodity servers within NFV infrastructures for incremental monitoring. For a network operator, deploying a set of vDPIs over the network is a matter of finding the appropriate placement that meets the traffic management and operational cost constraints (license fees, network efficiency or power consumption). In this paper, we formulate the vDPI placement problem as a cost minimization problem. We cast the problem as a multi-commodity flow problem. We then propose a centrality-based greedy algorithm and assess its validity by comparing it with the ILP optimal solution on random networks.
65 citations
22 May 2017
TL;DR: A quick guide to understand the complexity of the classifier evaluation process including the choice of evaluating metrics (scores) as well as the statistical comparison of classifiers.
Abstract: Performance evaluation of supervised classification learning method related to its prediction ability on independent data is very important in machine learning. It is also almost unthinkable to carry out any research work without the comparison of the new, proposed classifier with other already existing ones. This paper aims to review the most important aspects of the classifier evaluation process including the choice of evaluating metrics (scores) as well as the statistical comparison of classifiers. Critical view, recommendations and limitations of the reviewed methods are presented. The article provides a quick guide to understand the complexity of the classifier evaluation process and tries to warn the reader about the wrong habits.
50 citations
1 Jan 2013
TL;DR: Experiments indicate that this proposal of Local-and-Over-All Balanced bagging where probability of sampling an example is tuned according to the class distribution inside its neighbourhood is competitive to best undersampling bagging extensions.
Abstract: Various modifications of bagging for class imbalanced data are discussed. An experimental comparison of known bagging modifications shows that integrating with undersampling is more powerful than oversampling. We introduce Local-and-Over-All Balanced bagging where probability of sampling an example is tuned according to the class distribution inside its neighbourhood. Experiments indicate that this proposal is competitive to best undersampling bagging extensions.
47 citations
1 Jan 2013
TL;DR: A modified version of genetic algorithm is presented, capable of choosing feature sets of a slightly lower classification precision but significantly smaller number of features, which is presented via a benchmark EEG set submitted to the second BCI Competition.
Abstract: The crucial problem which has to be solved when an effective brain-computer interface (BCI) is to be design is: how to reduce the huge space of features extracted from raw EEG signals? One of the techniques of feature selection often used by BCI researches are genetic algorithms (GA). This approach, in its classic form, allows obtaining a feature set which gives the high classification precision, however, the dimension of this set is often still too large to create a reliable classifier. The paper presents a modified version of genetic algorithm, which is capable of choosing feature sets of a slightly lower classification precision but significantly smaller number of features. The practical application of the proposed algorithm will be presented via a benchmark EEG set submitted to the second BCI Competition (data set III - motor imaginary).
40 citations
Performance Metrics
| Year | Papers |
|---|---|
| 2019 | 39 |
| 2017 | 52 |
| 2016 | 86 |
| 2015 | 2 |
| 2013 | 86 |
| 2011 | 76 |