The specification and verification of algorithms is vital for safety-critical autonomous systems which incorporate deep learning elements. We propose an integrated process for verifying artificial neural network (ANN) classifiers. This process consists of an off-line verification and an on-line performance prediction phase. The process is intended to verify ANN classifier generalisation performance, and to this end makes use of dataset dissimilarity measures. We introduce a novel measure for quantifying the dissimilarity between the dataset used to train a classification algorithm, and the test dataset used to evaluate and verify classifier performance. A system-level requirement could specify the permitted form of the functional relationship between classifier performance and a dissimilarity measure; such a requirement could be verified by dynamic testing. Experimental results, obtained using publicly available datasets, suggest that the measures have relevance to real-world practice for both quantifying dataset dissimilarity, and specifying and verifying classifier performance.<\/p>","DOI":"10.4018\/ijaiml.289536","type":"journal-article","created":{"date-parts":[[2021,9,29]],"date-time":"2021-09-29T19:02:17Z","timestamp":1632942137000},"page":"1-21","source":"Crossref","is-referenced-by-count":2,"title":["An Integrated Process for Verifying Deep Learning Classifiers Using Dataset Dissimilarity Measures"],"prefix":"10.4018","volume":"11","author":[{"given":"Darryl","family":"Hond","sequence":"first","affiliation":[{"name":"Thales, UK"}]},{"ORCID":"http:\/\/orcid.org\/0000-0002-9317-7045","authenticated-orcid":true,"given":"Hamid","family":"Asgari","sequence":"additional","affiliation":[{"name":"Thales, UK"}]},{"given":"Daniel","family":"Jeffery","sequence":"additional","affiliation":[{"name":"Thales, UK"}]},{"given":"Mike","family":"Newman","sequence":"additional","affiliation":[{"name":"Thales, UK"}]}],"member":"2432","reference":[{"key":"IJAIML.289536-0","first-page":"252","article-title":"Classification of heterogeneous data based on data type impact on similarity.","author":"N.Ali","year":"2018","journal-title":"UK Workshop on Computational Intelligence"},{"key":"IJAIML.289536-1","unstructured":"Asgari, H., Farrell, J., & Pritchard, B. (2019). Review of regulatory issues of robotic and autonomous systems: Learning for civil nuclear industry. Engineering Safe Autonomy, 27th of Safety Critical Systems Symposium, Bristol, UK."},{"key":"IJAIML.289536-2","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58920-2_13"},{"key":"IJAIML.289536-3","unstructured":"Chico State\u2019s Software Engineering. (2021). Software testing - edge and corner cases. https:\/\/github.com\/ChicoState\/SoftwareEngineering\/wiki\/Software-Testing-Vocabulary"},{"key":"IJAIML.289536-4","unstructured":"Chung, Y., Haas, P. J., Upfal, E., & Kraska, T. (2018). Unknown examples & machine learning model generalization. arXiv preprint arXiv:1808.08294."},{"key":"IJAIML.289536-5","unstructured":"Geirhos, R., Janssen, D. H., Sch\u00fctt, H. H., Rauber, J., Bethge, M., & Wichmann, F. A. (2017). Comparing deep neural networks against humans: object recognition when the signal gets weaker. arXiv preprint arXiv:1706.06969"},{"issue":"1","key":"IJAIML.289536-6","first-page":"723","article-title":"A kernel two-sample test.","volume":"13","author":"A.Gretton","year":"2012","journal-title":"Journal of Machine Learning Research"},{"key":"IJAIML.289536-7","unstructured":"Hawkins, R. (2019). Body of knowledge - structure and scope. https:\/\/www.york.ac.uk\/assuring-autonomy\/"},{"key":"IJAIML.289536-8","doi-asserted-by":"publisher","DOI":"10.1109\/ICMLA51294.2020.00027"},{"key":"IJAIML.289536-9","unstructured":"Hond, D., White, A., & Asgari, H. (2020). Quantifying dataset properties for systematic artificial neural network classifier verification. Assuring Safe Autonomy, 28th of Safety Critical Systems Symposium, York, UK."},{"key":"IJAIML.289536-10","doi-asserted-by":"publisher","DOI":"10.1109\/ICSE.2019.00108"},{"key":"IJAIML.289536-11","unstructured":"Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980."},{"key":"IJAIML.289536-12","unstructured":"Kouw, W. M., & Loog, M. (2018). An introduction to domain adaptation and transfer learning. arXiv preprint arXiv:1812.11806."},{"key":"IJAIML.289536-13","unstructured":"LeCun, Y. (1998). The MNIST database of handwritten digits. http:\/\/yann.lecun.com\/exdb\/mnist\/"},{"key":"IJAIML.289536-14","doi-asserted-by":"publisher","DOI":"10.1109\/5.726791"},{"key":"IJAIML.289536-15","unstructured":"Liang, S., Li, Y., & Srikant, R. (2017). Enhancing the reliability of out-of-distribution image detection in neural networks. arXiv preprint arXiv:1706.02690."},{"key":"IJAIML.289536-16","doi-asserted-by":"publisher","DOI":"10.1145\/3238147.3238202"},{"key":"IJAIML.289536-17","unstructured":"Papernot, N., & McDaniel, P. (2018). Deep k-nearest neighbors: Towards confident, interpretable and robust deep learning. arXiv preprint arXiv:1803.04765."},{"key":"IJAIML.289536-18","doi-asserted-by":"publisher","DOI":"10.1145\/3132747.3132785"},{"key":"IJAIML.289536-19","unstructured":"Seo, J., Jeon, S., & Jeon, T. (2018). Domain adaptive generation of aircraft on satellite imagery via simulated and unsupervised learning. arXiv preprint arXiv:1806.03002."},{"key":"IJAIML.289536-20","doi-asserted-by":"publisher","DOI":"10.1016\/j.jss.2020.110574"},{"key":"IJAIML.289536-21","unstructured":"Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I., & Fergus, R. (2013). Intriguing properties of neural networks. arXiv preprint arXiv:1312.6199."},{"key":"IJAIML.289536-22","doi-asserted-by":"publisher","DOI":"10.1145\/3180155.3180220"},{"issue":"2","key":"IJAIML.289536-23","first-page":"58","article-title":"The problem of concept drift: Definitions and related work. Computer Science Department","volume":"106","author":"A.Tsymbal","year":"2004","journal-title":"Trinity College Dublin"},{"key":"IJAIML.289536-24","unstructured":"Tzeng, E., Hoffman, J., Zhang, N., Saenko, K., & Darrell, T. (2014). Deep domain confusion: Maximizing for domain invariance. arXiv preprint arXiv:1412.3474."},{"key":"IJAIML.289536-25","doi-asserted-by":"publisher","DOI":"10.1109\/72.788640"},{"key":"IJAIML.289536-26","unstructured":"Weng, T. W., Zhang, H., Chen, P. Y., Yi, J., Su, D., Gao, Y., Hsieh, C. J., & Daniel, L. (2018). Evaluating the robustness of neural networks: An extreme value theory approach. arXiv preprint arXiv:1801.10578."},{"key":"IJAIML.289536-27","unstructured":"Xiao, H., Rasul, K., & Vollgraf, R. (2017). Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747."}],"container-title":["International Journal of Artificial Intelligence and Machine Learning"],"original-title":[],"language":"ng","link":[{"URL":"https:\/\/www.igi-global.com\/viewtitle.aspx?TitleId=289536","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2023,1,16]],"date-time":"2023-01-16T22:42:11Z","timestamp":1673908931000},"score":1,"resource":{"primary":{"URL":"https:\/\/services.igi-global.com\/resolvedoi\/resolve.aspx?doi=10.4018\/IJAIML.289536"}},"subtitle":[""],"short-title":[],"issued":{"date-parts":[[2021,10,8]]},"references-count":28,"journal-issue":{"issue":"2","published-print":{"date-parts":[[2021,7]]}},"URL":"https:\/\/doi.org\/10.4018\/ijaiml.289536","relation":{},"ISSN":["2642-1577","2642-1585"],"issn-type":[{"value":"2642-1577","type":"print"},{"value":"2642-1585","type":"electronic"}],"subject":[],"published":{"date-parts":[[2021,10,8]]}}}