| 1. |
- Andersson, Robin, et al.
(författare)
-
RoSy : A Rough Knowledge Base System
- 2005
-
Ingår i: Rough Sets, Fuzzy Sets, Data Mining, and Granular-Soft Computing,2005. - Berlin : Springer. ; , s. 48-
-
Konferensbidrag (refereegranskat)
|
|
| 2. |
- Andersson, Robin, et al.
(författare)
-
RoSy: A Rough Knowledge Base System
- 2005
-
Ingår i: Rough Sets, Fuzzy Sets, Data Mining, and Granular Computing. - Berlin, Heidelberg : Springer Berlin Heidelberg. - 9783540286608 ; , s. 48-58
-
Konferensbidrag (refereegranskat)abstract
- This paper presents a user-oriented view of RoSy, a Rough Knowledge Base System. The system tackles two problems not fully answered by previous research: the ability to define rough sets in terms of other rough sets and incorporation of domain or expert knowledge. We describe two main components of RoSy: knowledge base creation and query answering. The former allows the user to create a knowledge base of rough concepts and checks that the definitions do not cause what we will call a model failure. The latter gives the user a possibility to query rough concepts defined in the knowledge base. The features of RoSy are described using examples. The system is currently available on a web site for online interactions.
|
|
| 3. |
- Baltzer, Nicholas, et al.
(författare)
-
Stratifying Cervical Cancer Risk With Registry Data
- 2018
-
Ingår i: 2018 IEEE 14th International Conference on e-Science (e-Science 2018). - : IEEE. - 9781538691564 ; , s. 288-289
-
Konferensbidrag (refereegranskat)abstract
- The cervical cancer screening programmes in Sweden and Norway have successfully reduced the frequency of cervical cancer incidence but have not implemented any form of evaluation for screening needs. This means that the screening frequency for individuals can he suboptimal, increasing either the cost of the programme or the risk of missing an early stage cancer development. We developed a framework for assessing an individual's risk of cervical cancer based on their available screening history and computing a primary risk factor called CRS from a data-driven separation model together with multiple derived attributes. The results show that this approach is highly practical, validates against multiple established trends, and can he effective in personalizing the screening needs for individuals.
|
|
| 4. |
|
|
| 5. |
|
|
| 6. |
- Bornelöv, Susanne, et al.
(författare)
-
Visualization of Rules in Rule-Based Classifiers
- 2012
-
Ingår i: INTELLIGENT DECISION TECHNOLOGIES (IDT'2012), VOL 1. - Berlin, Heidelberg : Springer Berlin Heidelberg. - 9783642299773 ; , s. 329-338
-
Konferensbidrag (refereegranskat)abstract
- Interpretation and visualization of the classification models are important parts of machine learning. Rule-based classifiers often contain too many rules to be easily interpreted by humans, and methods for post-classification analysis of the rules are needed. Here we present a strategy for circular visualization of sets of classification rules. The Circos software was used to generate graphs showing all pairs of conditions that were present in the rules as edges inside a circle. We showed using simulated data that all two-way interactions in the data were found by the classifier and displayed in the graph, although the single attributes were constructed to have no correlation to the decision class. For all examples we used rules trained using the rough set theory, but the visualization would by applicable to any sort of classification rules. This method for rule visualization may be useful for applications where interaction terms are expected, and the size of the model limits the interpretability.
|
|
| 7. |
- Kierczak, Marcin, et al.
(författare)
-
Construction of Rough Set-Based Classifiers for Predicting HIV Resistance to Nucleoside Reverse Transcriptase Inhibitors
- 2008
-
Ingår i: GRANULAR COMPUTING. - Berlin, Heidelberg : Springer Berlin Heidelberg. - 9783540769729 ; , s. 249-258
-
Konferensbidrag (refereegranskat)abstract
- For more than two decades AIDS remains a terminal disease and no efficient therapy exists. The high mutability of HIV leads to serious problems in designing efficient anti-viral drugs. Soon after introducing a new drug, there appear HIV strains that are resistant to the applied agent. In order to help overcoming resistance, we constructed a classificatory model of genotype-resistance relationship. To derive our model, we use rough sets theory. Furthermore, by incorporating existing biochemical knowledge into our model, it gains biological meaning and becomes helpful in understanding drug resistance phenomenon. Our highly accurate classifiers are based on a number of explicit, easy-to-interpret IF-THEN rules. For every position in amino acid sequence of viral enzyme reverse transcriptase (one of two main targets for anti-viral drugs), the rules describe the way the biochemical properties of amino acid have to change in order to acquire drug resistance. Preliminary biomolecular analysis suggests the applicability of the model.
|
|
| 8. |
- Strömbergsson, Helena, et al.
(författare)
-
Proteochemometrics modelling of receptor ligand interactions using rough sets
- 2004
-
Ingår i: Proceedings of the German conference on Bioinformatics. - 3885793822 ; , s. 85-94
-
Konferensbidrag (refereegranskat)abstract
- We report on a model for the interaction of chimeric melanocortin G-protein coupled receptors with peptide ligands using the rough set approach. Rough sets generate If-Then rule models using Boolean reasoning. Two separate datasets have been analyzed, for which the binding affinities have previously been measured experimentally. The receptors and ligands are described by vectors of strings. Different partitions of each dataset were evaluated in order to find an optimal partition into rough set decision classes. To obtain a measurement of the accuracy of the rough set classifier generated from each dataset, a 10-fold cross validation (CV) was performed. The Area Under Curve (AUC) was calculated for each iteration during CV. This resulted in an AUC mean of 0.94 (SD 0.12) and 0.93 (SD 0.16) for the first and second dataset respectively. The CV results show that the rough set models exhibit a high classification quality. The decision rules generated from the rough set model inductions are easy to interpret. We apply this information to develop models of the interaction between ligands and receptors.
|
|
