Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Matters Arising
  • Published:

Limitations in odour recognition and generalization in a neuromorphic olfactory circuit

Nature Machine Intelligence volume 6, pages 1451–1453 (2024)Cite this article

Subjects

Matters Arising to this article was published on 16 December 2024

The Original Article was published on 16 March 2020

Neuromorphic computing is one of the few current approaches that have the potential to substantially reduce power consumption in machine learning and artificial intelligence, and has drawn vast inspiration from considerations of biological systems and circuits. In their work, Imam and Cleland1 presented a neuromorphic odour-learning algorithm that is inspired by mammalian olfactory bulb circuitry, which they assessed by considering its performance in ‘rapid online learning and identification’ and ‘broad generalization beyond experience’ of gaseous odourants and odourless gases (in short, ‘gases’) using a set of gas-sensor recordings of different odour presentations and corrupting them by impulse noise. We replicated parts of the study and discovered limitations thereof, which are (1) that the dataset used suffers from sensor drift and a non-randomized measurement protocol that render it of limited use for odour identification benchmarks, and (2) that the model is restricted in its ability to generalize over repeated presentations of the same gas. Therefore, a validation of the model that goes beyond restoring a previously learned data sample remains to be shown, in particular its coherence with the attributed capabilities of robustness and broad generalization beyond experience, as well as its suitability to realistic odour identification tasks.

In addition to the dataset’s limitations, we found restrictions in the model’s capability to generalize over different recordings of the same stimulus. Generalization is an important property of any pattern-recognition system6. The authors convincingly show that the model can restore input patterns corrupted by impulse noise. However, in most instances, the authors tested recognition on the same sample that was used for training, occluding 60% of the sample with noise. Forty per cent of each training sample was present unchanged in the corresponding testing sample. By using overlapping portions of data for both training and testing, the statistical independence required for a robust assessment of generalization is not maintained. A real odourant recognition and signal restoration system would rarely encounter the exact same stimulus twice, once in a clean and once in a corrupted version. Therefore, assessing the model’s capability to recognize and restore patterns from separate recordings is essential to judge its relevance.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Overview of the data collection protocol.
Fig. 2: Ablation study, evaluating the EPL network under a range of conditions.

Data availability

The data used for our experiments are publicly available and described in ref. 2.

Code availability

Our experiments were based on the code released by the authors with the original study. Our adaptations and instructions for replication, together with the data, are available at https://github.com/BioMachineLearning/EPLNetworkImamCleland2020.

References

  1. Imam, N. & Cleland, T. A. Rapid online learning and robust recall in a neuromorphic olfactory circuit. Nat. Mach. Intell. 2, 181–191 (2020).

    Article  Google Scholar 

  2. Vergara, A. et al. On the performance of gas sensor arrays in open sampling systems using inhibitory support vector machines. Sens. Actuators B 185, 462–477 (2013).

    Article  Google Scholar 

  3. Vergara, A. et al. Chemical gas sensor drift compensation using classifier ensembles. Sens. Actuators B 166–167, 320–329 (2012).

    Article  Google Scholar 

  4. Dennler, N., Rastogi, S., Fonollosa, J., van Schaik, A. & Schmuker, M. Drift in a popular metal oxide sensor dataset reveals limitations for gas classification benchmarks. Sens. Actuators B 361, 131668 (2022).

    Article  Google Scholar 

  5. Hines, E. L., Llobet, E. & Gardner, J. Electronic noses: a review of signal processing techniques. IEE Proc. Circuits Devices Syst. 146, 297–310 (1999).

    Article  Google Scholar 

  6. Gareth, J., Daniela, W., Trevor, H. & Robert, T. An Introduction to Statistical Learning: with Applications in R (Springer, 2013); https://doi.org/10.1007/978-1-4614-7138-7

Download references

Acknowledgements

We thank N. Imam and T. A. Cleland for their valuable feedback and suggestions. We thank D. Drix, M. Psarrou, S. Rastogi and S. Sutton for discussions. We acknowledge the Telluride and CapoCaccia neuromorphic workshops and their participants. M.S. was funded from EU H2020 Grant Human Brain Project SGA3 (945539). This project is supported by the NSF/CIHR/DFG/FRQ/UKRI-MRC Next Generation Networks for Neuroscience Program (NSF 2014217/MRC MR/T046759/1’Odor2Action’).

Author information

Authors and Affiliations

  1. Biocomputation Group, University of Hertfordshire, Hatfield, UK

    Nik Dennler & Michael Schmuker

  2. International Centre for Neuromorphic Systems, The MARCS Institute for Brain, Behaviour, and Development, Western Sydney University, Kingswood, New South Wales, Australia

    Nik Dennler & André van Schaik

Authors
  1. Nik Dennler
    View author publications

    Search author on:PubMed Google Scholar

  2. André van Schaik
    View author publications

    Search author on:PubMed Google Scholar

  3. Michael Schmuker
    View author publications

    Search author on:PubMed Google Scholar

Contributions

N.D.: conceptualization, investigation, formal analysis, software, visualization, writing—original draft, and writing—review and editing. A.v.S: conceptualization, writing—review and editing, and supervision. M.S.: conceptualization, writing—review and editing, funding acquisition, and supervision.

Corresponding authors

Correspondence to Nik Dennler, André van Schaik or Michael Schmuker.

Ethics declarations

Competing interest

The authors declare no competing interests.

Peer review

Peer review information

Nature Machine Intelligence thanks Yang Chai for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

arising from N. Imam & T. A. Cleland Nature Machine Intelligence https://doi.org/10.1038/s42256-020-0159-4 (2020)

Supplementary information

Supplementary Information

Supplementary Figs. 1 and 2 and Algorithms 1 and 2.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Dennler, N., van Schaik, A. & Schmuker, M. Limitations in odour recognition and generalization in a neuromorphic olfactory circuit. Nat Mach Intell 6, 1451–1453 (2024). https://doi.org/10.1038/s42256-024-00952-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • Issue date:

  • DOI: https://doi.org/10.1038/s42256-024-00952-1

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing