Abstract
Odor classification by a robot equipped with an electronic nose (e-nose) is a challenging task for pattern recognition since volatiles have to be classified quickly and reliably even in the case of short measurement sequences, gathered under operation in the field. Signals obtained in these circumstances are characterized by a high-dimensionality, which limits the use of classical classification techniques based on unsupervised and semi-supervised settings, and where predictive variables can be only identified using wrapper or post-processing techniques. In this paper, we consider generative topographic mapping through time (GTM-TT) as an unsupervised model for time-series inspection, based on hidden Markov models regularized by topographic constraints. We further extend the model such that supervised classification and relevance learning can be integrated, resulting in supervised GTM-TT. Then, we evaluate the suitability of this new technique for the odor classification problem in robotics applications. The performance is compared with classical techniques as nearest neighbor, as an absolute baseline, support vector machine and a recent time-series kernel approach, demonstrating the eligibility of our approach for high-dimensional data. Additionally, we exploit the learning system introduced in this work, providing a measure of the relevance of each sensor and individual time points in the classification process, from which important information can be extracted.
Similar content being viewed by others
Notes
Figaro engineering inc. http://www.figaro.co.jp.
e2v. http://www.e2v.com/.
Grids: \({\lambda ,\gamma } = [0, 10^{-6} \ldots 10^{-1}, 0.5, 1 \ldots 5, 10, 30, 50, 100]\) \({\rm costs} = [0.1, 10, 10^2, 5 \times 10^2, 10^3, 5 \times 10^3, 10^4, 5 \times 10^4]\).
Approaches for feature ranking by SVM are available but not for this type of data and not directly for multi-class problems as studied for DS1.
Since butane is found at gas state at ambient temperature, the content of a lighter was released when the e-nose aspiration moved over the container.
Here we simply used the model from the first crossvalidation run.
References
Astudillo CA, Oommen BJ (2014) Topology-oriented self-organizing maps: a survey. Pattern Anal Appl 17(2):1–26
Bishop CM, Svensén M, Williams CKI (1998) Gtm: the generative topographic mapping. Neural Comput 10(1):215–234
Bishop CM (2006) Pattern recognition and machine learning. Information science and statistics. Springer, New York
Blanco J-L, Monroy JG González-Jiménez, J, Lilienthal A (2013) A Kalman filter based approach to probabilistic gas distribution mapping. In: 28th symposium on applied computing (SAC), March 2013.
Brahim-Belhouari S, Bermak A (2005) Gas identification using density models. Pattern Recognit Lett 26(6):699–706
Chen H, Tang F, Tino P, Yao X (2013) Model-based kernel for efficient time-series analysis. In: Proceedings of the 19th ACM SIGKDD conference on knowledge discovery and data mining (KDD) (KDD’13), Chicago, USA
Costa IG, Schönhuth A, Hafemeister C, Schliep A (2009) Constrained mixture estimation for analysis and robust classification of clinical time-series. Bioinformatics 25(12):i6–i14
Distante C, Siciliano P, Persaud KC (2002) Dynamic cluster recognition with multiple self-organising maps. Pattern Anal Appl 5(3):306–315
Distante C, Leo M, Siciliano P, Persaud KC (2002) On the study of feature extraction methods for an electronic nose. Sens Actuators B Chem 87(2):274–288
Douzal-Chouakria A, Amblard C (2012) Classification trees for time-series. Pattern Recognit 45(3):1076–1091
Dumitrescu D, Lazzerini B, Marcelloni F (2000) A fuzzy hierarchical classification system for olfactory signals. Pattern Anal Appl 3(4):325–334
Ehret B, Safenreiter K, Lorenz F, Biermann J (2011) A new feature extraction method for odour classification. Sens Actuators B Chem 158(1):75–88
Ferri G, Caselli E, Mattoli V, Mondini A, Mazzolai B, Dario P (2009) Spiral: anovel biologically-inspired algorithm for gas/odor source localization in an indoor environment with no strong airflow. Robot Auton Sys 57(4):393–402
Fonollosa J, Rodrguez-Lujn I, Trincavelli M, Vergara A, Huerta R (2014) Chemical discrimination in turbulent gas mixtures with mox sensors validated by gas chromatography-mass spectrometry. Sensors 14(10):19336–19353
Gisbrecht A, Hammer B (2011) Relevance learning in generative topographic mapping. Neurocomputing 74(9):1359–1371
González-Jiménez J,Monroy JG, Blanco J-L (2013) Robots that can smell: motivation and problems. In: Submitted to 15th international symposium on olfaction and electronic nose (ISOEN)
Gonzalez-Jimenez J, Monroy JG, Blanco J-L (2011) The multi-chamber electronic nose. An improved olfaction sensor for mobile robotics. Sensors 11(6):6145–6164
Gutierrez-Osuna R (2002) Pattern analysis for machine olfaction: a review. Sens J IEEE 2(3):189–202
Hammer B, Villmann TH (2002) Generalized relevance learning vector quantization. Neural Netw 15(8–9):1059–1068
Hatami N, Chira C (2013) Classifiers with a reject option for early time-series classification. In: Proceedings of the IEEE symposium on computational intelligence and ensemble learning, CIEL 2013, IEEE symposium series on computational intelligence (SSCI), 16–19 April 2013. IEEE, Singapore, pp 9–16
Lee J, Verleysen M (2005) Generalizations of the lp norm for time-series and its application to self-organizing maps. In: Cottrell M (ed) 5th workshop on self-organizing maps, vol 1, pp 733–740
Lilienthal AJ, Reggente M, Trincavelli M, Blanco J-L, Gonzalez J (2009) A statistical approach to gas distribution modelling with mobile robots—the kernel dmv algorithm. In: Intelligent robots and systems, 2009. IROS 2009. IEEE/RSJ international conference, pp 570–576
Lin T-H, Kaminski N, Bar-Joseph Z (2008) Alignment and classification of time-series gene expression in clinical studies. In: ISMB, pp 147–155
Lisboa PJG (2013) Interpretability in machine learning—principles and practice. In: Masulli F, Pasi G, Yager RR (eds) Fuzzy logic and applications—10th international workshop, WILF 2013, Genoa, Italy, 19–22 Nov 2013. Proceedings, Lecture Notes in Computer Science, vol 8256. Springer, pp 15–21
Loutfi A, Coradeschi S, Lilienthal AJ, Gonzalez J (2009) Gas distribution mapping of multiple odour sources using a mobile robot. Robotica 27(2):311–319
Marques L, Nunes, U, de Almeida AT (2002) Olfaction-based mobile robot navigation. Thin Solid Films, 418(1):51–58. Proceedings from the international school on gas sensors in conjunction with the 3rd European school of the NOSE network
Martinelli E, Magna G, Vergara A, Di Natale C (2014) Cooperative classifiers for reconfigurable sensor arrays. Sens Actuators B Chem 199:83–92
Michalak M (2011) Adaptive kernel approach to the time-series prediction. Pattern Anal Appl 14(3):283–293
Monroy JG, Gonzlez-Jimnez J, Blanco J-L (2012) Overcoming the slow recovery of mox gas sensors through a system modeling approach. Sensors 12(10):13664–13680
Moseley P, Tofield B (eds) (1987) Solid state gas sensors. Adam Hilger, Bristol
Olier I, Vellido A (2008) Advances in clustering and visualization of time-series using gtm through time. Neural Netw 21(7):904–913
Petitjean F, Ketterlin A, Ganarski P (2011) A global averaging method for dynamic time warping, with applications to clustering. Pattern Recognit 44(3):678–693
Prieto OJ, Alonso-González CJ, Rodríguez JJ (2013) Stacking for multivariate time-series classification. Pattern Anal Appl 1–16. doi:10.1007/s10044-013-0351-9
Robotics K. The jaco research edition robotic arm. http://www.kinovarobotics.com/
Sauve AC, Speed TP (2004) Normalization, baseline correction and alignment of high-throughput mass spectrometry data. In: Proceedings of Gensips. http://www.stat.berkeley.edu/~terry/Group/publications.html
Schleif F-M, Gisbrecht A, Hammer B (2012) Relevance learning for short high-dimensional time-series in the life sciences. Proceedings of IJCNN, pp 2069–2076
Schleif F-M, Mokbel B, Gisbrecht A, Theunissen L, Dürr V, Hammer B (2012) Learning relevant time points for time-series data in the life sciences. In: Proceedings of ICANN, pp 531–539
Schleif F-M, Ongyerth F-M, Villmann T (2009) Supervised data analysis and reliability estimation for spectral data. Neurocomputing 72(16–18):3590–3601
Schneider P, Biehl M, Hammer B (2009) Distance learning in discriminative vector quantization. Neural Comput 21:2942–2969
Shaffer RE, Rose-Pehrsson SL, Andrew McGill R (1999) A comparison study of chemical sensor array pattern recognition algorithms. Anal Chim Acta 384(3):305–317
Shumway RH, Stoffer DS (2000) Time-series analysis and Its applications. Springer Texts In Statistics. Springer, New York
Strachan IGD (2002) Latent variable methods for visualization through time. Ph.D. thesis, University of Edinburgh, Edinburgh
Strickert M, Hammer B (2005) Merge SOM for temporal data. Neurocomputing 64:39–72
Szczurek A, Krawczyk B, Maciejewska M (2013) VOCs classification based on the committee of classifiers coupled with single sensor signals. Chemometr Intell Lab Syst 125:1–10
Trincavelli M, Coradeschi S, Loutfi A (2009) Odour classification system for continuous monitoring applications. Sens Actuators B Chem 58:265–273
Vergara A, Fonollosa J, Mahiques J, Trincavelli M, Rulkov N, Huerta R (2013) On the performance of gas sensor arrays in open sampling systems using inhibitory support vector machines. Sens Actuators B Chem 185:462–477
Vergara A, Vembu S, Ayhan T, Ryan MA, Homer ML, Huerta R (2012) Chemical gas sensor drift compensation using classifier ensembles. Sens Actuators B Chem 166–167:320–329
Welch LR (2003) Hidden Markov models and the Baum–Welch algorithm. IEEE Inf Theory Soc Newsl 53(4). http://www.itsoc.org/publications/nltr/it\_dec\_03final.pdf
Ziyatdinov A, Fonollosa J, Fernndez L, Gutierrez-Glvez A, Marco S, Perera A (2015) Bioinspired early detection through gas flow modulation in chemo-sensory systems. Sens Actuators B Chem 206:538–547
Acknowledgments
The first author likes to thank Peter Tino, University of Birmingham, for interesting discussions about probabilistic modeling and Tien-ho Lin, Carnegie Mellon University, USA for support with the simulation data. Further, Ivan Olier, University of Manchaster, UK; Iain Strachan, AEA Technology, Harwell, UK and Markus Svensen, Microsoft Research, Cambridge, UK for providing code for GTM and GTM-TT. Further we would like to thank Fengzhen Tang, University of Birmingham for providing invaluable support with the RTK method. This work was supported by the DFG project HA2719/4-1 to BH, by the DFG-NSF project TO 409/8-1, and by the Cluster of Excellence 277 CITEC funded in the framework of the German Excellence Initiative. Further, a Marie Curie Intra-European Fellowship (IEF): FP7-PEOPLE-2012-IEF (FP7-327791-ProMoS) is gratefully acknowledged. Additional support was provided by funds from the Andalucía Regional Government and the European Union (FEDER) under research project: TEP08-4016.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Schleif, FM., Hammer, B., Monroy, J.G. et al. Odor recognition in robotics applications by discriminative time-series modeling. Pattern Anal Applic 19, 207–220 (2016). https://doi.org/10.1007/s10044-014-0442-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10044-014-0442-2