The Influence of Multi-class Feature Selection on the Prediction of Diagnostic Phenotypes


In this work, we evaluate two schemes for incorporating feature selection processes in multi-class classifier systems on high-dimensional data of low cardinality. These schemes operate on the level of the systems’ individual base classifiers and therefore do not perfectly fit in the traditional categories of filter, wrapper and embedded feature selection strategies. They can be seen as two examples of feature selection networks that are only loosely related to the structure of the multi-class classifier system. The architectures are tested for their application in predicting diagnostic phenotypes from gene expression profiles. Their selection stability and the overall generalization ability are evaluated in \(10 \times 10\) cross-validation experiments with support vector machines, random forests and nearest neighbor classifiers on eight publicly available multi-class microarray datasets. Overall the feature selecting multi-class classifier systems were able to outperform their counterparts on at least five of eight datasets.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10


  1. 1.

    Alon U, Barkai N, Notterman DA, Gish K, Ybarra S, Mack D, Levine AJ (1999) Broad patterns of gene expression revealed by clustering analysis of tumor and normal colon tissues probed by oligonucleotide arrays. Proc Natl Acad Sci USA 96(12):6745–6750

    Article  Google Scholar 

  2. 2.

    Ben-Dor A, Bruhn L, Friedman N, Nachman I, Schummer M, Yakhini Z (2000) Tissue classification with gene expression profiles. J Comput Biol 7(3–4):559–583

    Article  Google Scholar 

  3. 3.

    Berchtold NC, Cribbs DH, Coleman PD, Rogers J, Head E, Kim R, Beach T, Miller C, Troncoso J, Trojanowski JQ, Zielke HR, Cotman CW (2008) Gene expression changes in the course of normal brain aging are sexually dimorphic. Proc Natl Acad Sci USA 105(40):15,605–15,610

    Article  Google Scholar 

  4. 4.

    Blum A, Langley P (1997) Selection of relevant features and examples in machine learning. Artif Intell 97(1–2):245–271

    MathSciNet  Article  Google Scholar 

  5. 5.

    Breiman L (2001) Random forests. Mach Learn 45(1):5–32

    Article  Google Scholar 

  6. 6.

    Chen R, Snyder M (2013) Promise of personalized omics to precision medicine. Wiley Interdiscip Rev Syst Biol Med 5(1):73–82

    Article  Google Scholar 

  7. 7.

    Cover TM (1965) Geometrical and statistical properties of systems of linear inequalities with applications in pattern recognition. IEEE Trans Electron Comput 14(3):326–334

    Article  Google Scholar 

  8. 8.

    Dietterich TG, Bariki G (1995) Solving multiclass problems via error-correcting output codes. J Artif Intell Res 2:263–286

    Article  Google Scholar 

  9. 9.

    Fix E, Hodges JL (1951) Discriminatory analysis: nonparametric discrimination: consistency properties. Tech. Rep. Project 21-49-004, Report Number 4, USAF School of Aviation Medicine, Randolf Field, Texas

  10. 10.

    Freund Y, Schapire RE (1995) A decision-theoretic generalization of on-line learning and an application to boosting. In: Vitányi P (ed) Computational learning theory. Lecture Notes in Artificial Intelligence, vol 904. Springer, Berlin, pp 23–37

    Google Scholar 

  11. 11.

    Gobble RM, Qin LX, Brill ER, Angeles CV, Ugras S, O’Connor RB, Moraco NH, DeCarolis PL, Antonescu C, Singer S (2011) Expression profiling of liposarcoma yields a multigene predictor of patient outcome and identifies genes that contribute to liposarcomagenesis. Cancer Res 71(7):2697–2705

    Article  Google Scholar 

  12. 12.

    Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD, Lander ES (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science 286(5439):531–537

    Article  Google Scholar 

  13. 13.

    Gress TM, Kestler HA, Lausser L, Fiedler L, Sipos B, Michalski CW, Werner J, Giese N, Scarpa A, Buchholz M (2011) Differentiation of multiple types of pancreatico-biliary tumors by molecular analysis of clinical specimens. J Mol Med 90(4):457–464

    Article  Google Scholar 

  14. 14.

    Guyon I, Elisseeff A (2003) An introduction to variable and feature selection. J Mach Learn Res 3:1157–1182

    MATH  Google Scholar 

  15. 15.

    Haferlach T, Kohlmann A, Wieczorek L, Basso G, Kronnie GT, Béné MC, Vos JD, Hernández JM, Hofmann WK, Mills KI, Gilkes A, Chiaretti S, Shurtleff SA, Kipps TJ, Rassenti LZ, Yeoh AE, Papenhausen PR, Liu WM, Williams PM, Foà R (2010) Clinical utility of microarray-based gene expression profiling in the diagnosis and subclassification of leukemia: report from the international microarray innovations in leukemia study group. J Clin Oncol 28(15):2529–2537

    Article  Google Scholar 

  16. 16.

    Hastie T, Tibshirani R, Friedman JH (2001) The elements of statistical learning. Springer series in statistics. Springer, New York

    Google Scholar 

  17. 17.

    Huang Y, Suen C (1995) A method of combining multiple experts for the recognition of unconstrained handwritten numerals. IEEE Trans Pattern Anal Mach Intell 17(1):90–94

    Article  Google Scholar 

  18. 18.

    Jameson J, Longo D (2015) Precision medicine—personalized, problematic, and promising. N Engl J Med 372(23):2229–2234

    Article  Google Scholar 

  19. 19.

    Japkowicz N, Shah M (2011) Evaluating learning algorithms: a classification perspective. Cambridge University Press, New York

    Google Scholar 

  20. 20.

    Jones J, Otu H, Spentzos D, Kolia S, Inan M, Beecken WD, Fellbaum C, Gu X, Joseph M, Pantuck AJ, Jonas D, Libermann TA (2005) Gene signatures of progression and metastasis in renal cell cancer. Clin Cancer Res 11(16):5730–5739

    Article  Google Scholar 

  21. 21.

    Khan J, Wei J, Ringner M, Saal L, Westermann F, Berthold F, Schwab M, Antonesco C, Peterson C, Meltzer P (2001) Classification and diagnostic prediction of cancer using gene expression profiling and artificial neural networks. Nat Med 7(6):673–679

    Article  Google Scholar 

  22. 22.

    Kimpel MW, Strother WN, McClintick JN, Carr LG, Liang T, Edenberg HJ, McBride WJ (2007) Functional gene expression differences between inbred alcohol-preferring and non-preferring rats in five brain regions. Alcohol 41(2):95–132

    Article  Google Scholar 

  23. 23.

    Kohavi R, John G (1997) Wrappers for feature subset selection. Artif Intell 97(1–2):273–324

    Article  Google Scholar 

  24. 24.

    Kuncheva LI (2004) Combining pattern classifiers: methods and algorithms. Wiley, Hoboken

    Google Scholar 

  25. 25.

    Lattke R, Lausser L, Müssel C, Kestler HA (2015) Detecting ordinal class structures. In: Schwenker F, Roli F, Kittler J (eds) Multiple classifier systems, MCS 2015, Lecture notes in computer science, vol 9132. Springer, Cham, pp 100–111

    Google Scholar 

  26. 26.

    Lausser L, Kestler HA (2010) Robustness analysis of eleven linear classifiers in extremely high-dimensional feature spaces. In: Schwenker F, El Gayar N (eds) Artificial neural networks in pattern recognition. ANNPR 2010, Lecture Notes in Artificial Intelligence, vol 5998. Springer, Berlin, Heidelberg, pp 72–83

    Google Scholar 

  27. 27.

    Lausser L, Kestler HA (2014) Fold change classifiers for the analysis of gene expression profiles. In: Gaul W, Geyer-Schulz A, Baba Y, Okada A (eds) German–Japanese interchange of data analysis results. Studies in classification, data analysis, and knowledge organization. Springer, Cham, pp 193–202

    Google Scholar 

  28. 28.

    Lausser L, Müssel C, Kestler HA (2013) Measuring and visualizing the stability of biomarker selection techniques. Comput Stat 28(1):51–65

    MathSciNet  Article  Google Scholar 

  29. 29.

    Lorena AC, de Carvalho ACPLF, Gama JMP (2009) A review on the combination of binary classifiers in multiclass problems. Artif Intell Rev 30:19–37

    Article  Google Scholar 

  30. 30.

    Maire V, Baldeyron C, Richardson M, Tesson B, Salomon AV, Gravier E, Marty-Prouvost B, Koning LD, Rigaill G, Dumont A, Gentien D, Barillot E, Roman-Roman S, Depil S, Cruzalegui F, Pierré A, Tucker GC, Dubois T (2013) TTK/hMPS1 is an attractive therapeutic target for triple-negative breast cancer. PLoS One 8(5):e63712

    Article  Google Scholar 

  31. 31.

    Müssel C, Lausser L, Maucher M, Kestler HA (2012) Multi-objective parameter selection for classifiers. J Stat Softw 46(5):1–27

    Article  Google Scholar 

  32. 32.

    Palm G (2016) Neural information processing in cognition: we start to understand the orchestra, but where is the conductor? Front Comput Neurosci 10:1–6

    Article  Google Scholar 

  33. 33.

    Pfister TD, Reinhold WC, Agama K, Gupta S, Khin SA, Kinders RJ, Parchment RE, Tomaszewski JE, Doroshow JH, Pommier Y (2009) Topoisomerase I levels in the NCI-60 cancer cell line panel determined by validated ELISA and microarray analysis and correlation with indenoisoquinoline sensitivity. Mol Cancer Ther 8(7):1878–1884

    Article  Google Scholar 

  34. 34.

    Ripley BD (1996) Pattern recognition and neural networks. Cambridge University Press, Cambridge

    Google Scholar 

  35. 35.

    Saeys Y, Iñza I, Larrañaga P (2007) A review of feature selection techniques in bioinformatics. Bioinformatics 23(19):2507–2517

    Article  Google Scholar 

  36. 36.

    Schirra LR, Lausser L, Kestler HA (2016a) Selection stability as a means of biomarker discovery in classification. In: Wilhelm AFX, Kestler HA (eds) Analysis of large and complex data. Studies in classification, data analysis, and knowledge organization. Springer, Cham, pp 79–89

    Google Scholar 

  37. 37.

    Schirra LR, Schmid F, Kestler HA, Lausser L (2016b) Interpretable classifiers in precision medicine: Feature selection and multi-class categorization. In: Schwenker F, Abbas HM, El Gayar N, Trentin E (eds) Artificial neural networks in pattern recognition, ANNPR 2016, Lecture Notes in Artificial Intelligence, vol 9896. Springer, pp 105–116

  38. 38.

    Skrzypczak M, Goryca K, Rubel T, Paziewska A, Mikula M, Jarosz D, Pachlewski J, Oledzki J, Ostrowski J (2010) Modeling oncogenic signaling in colon tumors by multidirectional analyses of microarray data directed for maximization of analytical reliability. PloS One 5(10):e13091

    Article  Google Scholar 

  39. 39.

    Taudien S, Lausser L, Giamarellos-Bourboulis EJ, Sponholz C, S F, Felder M, Schirra LR, Schmid F, Gogos C, G S, Petersen BS, Franke A, Lieb W, Huse K, Zipfel PF, Kurzai O, Moepps B, Gierschik P, Bauer M, Scherag A, Kestler HA, Platzer M (2016) Genetic factors of the disease course after sepsis: rare deleterious variants are predictive. EBioMedicine 12:227–238

    Article  Google Scholar 

  40. 40.

    Vapnik VN (1998) Statistical learning theory. Wiley, New York

    Google Scholar 

  41. 41.

    Webb AR (2002) Statistical pattern recognition, 2nd edn. Wiley, Chichester

    Google Scholar 

  42. 42.

    West M, Blanchette C, Dressman H, Huang E, Ishida S, Spang R, Zuzan H, Olson JAJ, Marks JR, Nevins JR (2001) Predicting the clinical status of human breast cancer by using gene expression profiles. Proc Natl Acad Sci USA 98(20):11462–11467

    Article  Google Scholar 

Download references


The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under Grant Agreement No. 602783, the German Research Foundation (DFG, SFB 1074 project Z1), and the Federal Ministry of Education and Research (BMBF, Gerontosys II, Forschungskern SyStaR, ID 0315894A and e:Med, SYMBOL-HF, ID 01ZX1407A) all to HAK.

Author information



Corresponding author

Correspondence to Hans A. Kestler.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (pdf 1326 KB)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lausser, L., Szekely, R., Schirra, L. et al. The Influence of Multi-class Feature Selection on the Prediction of Diagnostic Phenotypes. Neural Process Lett 48, 863–880 (2018).

Download citation


  • Multi-class classification
  • Classifier fusion
  • Feature selection
  • High-dimensional data
  • Low cardinality