Abstract
Machine learning is a popular method of learning functions from data to represent and to classify sensor inputs, multimedia, emails, and calendar events. Smartphone applications have been integrating more and more intelligence in the form of machine learning. Machine learning functionality now appears on most smartphones as voice recognition, spell checking, word disambiguation, face recognition, translation, spatial reasoning, and even natural language summarization. Excited app developers who want to use machine learning on mobile devices face one serious constraint that they did not face on desktop computers or cloud virtual machines: the end-user’s mobile device has limited battery life, thus computationally intensive tasks can harm end users’ phone availability by draining batteries of their stored energy. Currently, there are few guidelines for developers who want to employ machine learning on mobile devices yet are concerned about software energy consumption of their applications. In this paper, we combine empirical measurements of different machine learning algorithm implementations with complexity theory to provide concrete and theoretically grounded recommendations to developers who want to employ machine learning on smartphones. We conclude that some implementations of algorithms, such as J48, MLP, and SMO, do generally perform better than others in terms of energy consumption and accuracy, and that energy consumption is well-correlated to algorithmic complexity. However, to achieve optimal results a developer must consider their specific application as many factors — dataset size, number of data attributes, whether the model will require updating, etc. — affect which machine learning algorithm and implementation will provide the best results.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Weka for Android https://github.com/rjmarsan/Weka-for-Android
References
Abdulsalam H, Skillicorn DB, Martin P (2011) Classification using streaming random forests. IEEE Trans Knowl Data Eng 23(1):22–36
Aggarwal K, Hindle A, Stroulia E (2015a) Greenadvisor: A tool for analyzing the impact of software evolution on energy consumption. In: International Conference on Software Maintenance and Evolution (ICSME 2015), pp 311–320. http://softwareprocess.ca/pubs/aggarwal2015ICSME-greenadvisor.pdf
Aggarwal K, Hindle A, Stroulia E (2015b) Greenadvisor: a tool for analyzing the impact of software evolution on energy consumption. In: 31St IEEE international conference on software maintenance and evolution. IEEE computer society
Agolli T, Pollock L, Clause J (2017) Investigating decreasing energy usage in mobile apps via indistinguishable color changes. In: 2017 IEEE/ACM 4Th international conference on mobile software engineering and systems (MOBILESoft), pp 30–34. https://doi.org/10.1109/MOBILESoft.2017.17
Akdeniz: Google Play Crawler (2013) https://github.com/Akdeniz/google-play-crawler (Last accessed: May 2018)
Aha D, Kibler D (1991) Instance-based learning algorithms. Mach Learn 6:37–66
App Annie (2018a) App Annie. https://www.appannie.com/ (Last accessed: May 2018)
AppBrain (2018b) Top Android phones. http://www.appbrain.com/stats/top-android-phones (Last accessed: May 2018)
Apple Inc (2017) The future is here: iphone https://www.apple.com/newsroom/2017/09/the-future-is-here-iphone-x/ (Retrieved April 2018)
Banerjee A, Chong LK, Chattopadhyay S, Roychoudhury A (2014) Detecting energy bugs and hotspots in mobile apps. In: Proceedings of the 22nd ACM SIGSOFT International Symposium on Foundations of Software Engineering. ACM, pp 588–598
Benfield L (2018) Cfr - another java decompiler. http://www.benf.org/other/cfr/ (Last accessed: May 2018)
Bhattacharya S, Lane ND (2016) Sparsification and separation of deep learning layers for constrained resource inference on wearables. In: Proceedings of the 14th ACM Conference on Embedded Network Sensor Systems CD-ROM, SenSys ’16. ACM, New York, pp 176–189. https://doi.org/10.1145/2994551.2994564
Bottou L (1998) Online Algorithms and Stochastic Approximations. Cambridge University Press, Cambridge
Breiman L (2001) Random forests. Mach Learn 45(1):5–32
Bruce BR, Petke J, Harman M (2015) Reducing energy consumption using genetic improvement. In: Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation, GECCO ’15. ACM, New York, pp 1327–1334. https://doi.org/10.1145/2739480.2754752
Chenlei Z, Hindle A, German DM (2014) The impact of user choice on energy consumption. IEEE Software, pp 69–75. http://softwareprocess.ca/pubs/zhang2014IEEESoftware-user-choice.pdf
Chowdhury SA, Hindle A (2016a) Greenoracle: Estimating software energy consumption with energy measurement corpora. In: Proceedings of the 13th International Conference on Mining Software Repositories, MSR ’16. ACM, New York, pp 49–60. https://doi.org/10.1145/2901739.2901763
Chowdhury S, Sapra V, Hindle A (2016b) Client-side energy efficiency of http/2 for web and mobile app developers. In: 23rd IEEE International Conference on Software Analysis, Evolution, and Reengineering (SANER 2016), pp 529–540. https://doi.org/10.1109/SANER.2016.77. http://softwareprocess.ca/pubs/chowdhury2016SANER-http2.pdf
Christina B (2013) Your smartphone gains a mind of its own. Conde Nast http://www.wired.com/2013/07/ai-apps-trend/
Dex2jar download - Sourceforge.net (2018) http://sourceforge.net/projects/dex2jar/. (Last accessed: May 2018)
Di Nucci D, Palomba F, Prota A, Panichella A, Zaidman A, De Lucia A (2017) Software-based energy profiling of android apps: simple, efficient and reliable?. In: 2017 IEEE 24th international conference on Software analysis, evolution and reengineering (SANER). IEEE, pp 103–114
Frank E (2016) Class j48 http://weka.sourceforge.net/doc.dev/weka/classifiers/trees/J48.html
Good O (2015) How google translate squeezes deep learning onto a phone. Google Research Blog https://research.googleblog.com/2015/07/how-google-translate-squeezes-deep.html
Google (2016) Find time for your goals with google calendar. Google Blog https://googleblog.blogspot.ca/2016/04/find-time-goals-google-calendarhtml
Google (2017) Neural networks api: Android developers https://developer.android.com/ndk/guides/neuralnetworks/index.html (Last accessed: May 2018)
Google (2018) Mobile vision https://developers.google.com/vision/ (Last accessed: May 2018)
Greene T (2017) Google brings on-device machine learning to mobile with tensorflow lite https://thenextweb.com/artificial-intelligence/2017/11/15/google-brings-on-device-machine-learning-to-mobile-with-tensorflow-lite/ (Retrieved January 2018)
Gui J, Mcilroy S, Nagappan M, Halfond WGJ (2015) Truth in advertising: The hidden cost of mobile ads for software developers. In: 37Th IEEE/ACM international conference on software engineering, ICSE 2015, florence, Italy. IEEE, vol 1, pp 100–110. https://doi.org/10.1109/ICSE.2015.32
Gui J, Li D, Wan M, Halfond WG (2016) Lightweight measurement and estimation of mobile ad energy consumption. In: 2016 IEEE/ACM 5th international workshop on Green and sustainable software (GREENS). IEEE, pp 1–7
Hall M, Frank E, Holmes G, Pfahringer B, Reutemann P, Witten IH (2009) The weka data mining software: An update. SIGKDD Explor Newsl 11 (1):10–18. https://doi.org/10.1145/1656274.1656278
Hasan S, King Z, Hafiz M, Sayagh M, Adams B, Hindle A (2016) Energy profiles of java collections classes. In: International Conference on Software Engineering (ICSE 2016), pp 225–236. https://doi.org/10.1145/2884781.2884869 http://softwareprocess.ca/pubs/hasan2016ICSE-Energy-Profiles-of-Java-Collections-Classes.pdf
Hern Alex, A (2015) Smartphone now most popular way to browse internet – ofcom report. https://www.theguardian.com/technology/2015/aug/06/smartphones-most-popular-way-to-browse-internet-ofcom/ (last accessed: May 2018)
Hindle A, Ernst NA, Godfrey MW, Mylopoulos J (2013) Automated topic naming supporting cross-project analysis of software maintenance activities. J Empir Softw Eng 18(6):1125–1155. http://softwareprocess.ca/pubs/hindle2011EMSE-automated-topic-naming.pdf
Hindle A, Wilson A, Rasmussen K, Barlow EJ, Campbell J, Romansky S (2014) Greenminer: a hardware based mining software repositories software energy consumption framework. In: International Working Conference on Mining Software Repositories (MSR 2014), pp 12–21. http://softwareprocess.ca/pubs/hindle2014MSR-greenminer.pdf
Inc. A (2017) Core ml: Apple developer documentation https://developer.apple.com/documentation/coreml (Last accessed: May 2018)
Inc. W (2018) Wit.ai: Natural language for developers https://wit.ai/ (Last accessed: May 2018)
Platt J (1998) Fast training of support vector machines using sequential minimal optimization. In: Schoelkopf B, Burges C, Smola A (eds) Advances in Kernel Methods - Support Vector Learning. http://research.microsoft.com/~jplatt/smo.html. MIT Press
Jabbarvand R, Sadeghi A, Garcia J, Malek S, Ammann P (2015) Ecodroid: an approach for energy-based ranking of android apps. In: Proceedings of the Fourth International Workshop on Green and Sustainable Software. IEEE Press, pp 8–14
John GH, Langley P (1995) Estimating continuous distributions in bayesian classifiers. In: Eleventh conference on uncertainty in artificial intelligence. Morgan kaufmann, pp 338–345
Konradsson T (2015) Art and dalvik performance compared. Master’s thesis, UmeUniversity. http://www8.cs.umu.se/education/examina/Rapporter/TobiasKonradsson.pdf
LeCun Y, Cortes C, Burges CJ (1998) The mnist database of handwritten digits. http://yann.lecun.com/exdb/mnist/
Li D, Hao S, Gui J, Halfond WGJ (2014) An empirical study of the energy consumption of android applications. In: 30Th IEEE international conference on software maintenance and evolution, victoria, BC. IEEE Computer Society, pp 121–130. https://doi.org/10.1109/ICSME.2014.34
Li D, Lyu Y, Gui J, Halfond WGJ (2016) Automated energy optimization of http requests for mobile applications. In: Proceedings of the 38th International Conference on Software Engineering, ICSE ’16. ACM, New York, pp 249–260. https://doi.org/10.1145/2884781.2884867
Linares-Vásquez M, Bavota G, Bernal-Cárdenas C, Oliveto R, Di Penta M, Poshyvanyk D (2014) Mining energy-greedy api usage patterns in android apps: an empirical study. In: Proceedings of the 11th Working Conference on Mining Software Repositories, MSR 2014. ACM, New York, pp 2–11. https://doi.org/10.1145/2597073.2597085
Linares-Vásquez M, Bavota G, Cárdenas CEB, Oliveto R, Di Penta M, Poshyvanyk D (2015) Optimizing energy consumption of guis in android apps: a multi-objective approach. In: Proceedings of the 2015 10th Joint Meeting on Foundations of Software Engineering. ACM, New York, pp 143–154. https://doi.org/10.1145/2786805.2786847
Linares-Vásquez M, Vendome C, Tufano M, Poshyvanyk D (2017) How developers micro-optimize android apps. J Syst Softw 130:1–23
Lichman M (2013) UCI machine learning repository. http://archive.ics.uci.edu/ml
Machine Learning Laboratory (2015) Mnist arff files. http://axon.cs.byu.edu/data/mnist/
Malik H, Zhao P, Godfrey M (2015) Going green: An exploratory analysis of energy-related questions. In: Proceedings of the 12th Working Conference on Mining Software Repositories, MSR ’15. IEEE Press, Piscataway, pp 418–421. http://dl.acm.org/citation.cfm?id=2820518.2820576
Manotas I, Pollock L, Clause J (2014) Seeds: a software engineer’s energy-optimization decision support framework. In: Proceedings of the 36th International Conference on Software Engineering, ICSE 2014. ACM, New York, pp 503–514. https://doi.org/10.1145/2568225.2568297
Manotas I, Bird C, Zhang R, Shepherd D, Jaspan C, Sadowski C, Pollock L, Clause J (2016) An empirical study of practitioners’ perspectives on green software engineering. In: Proceedings of the 38th International Conference on Software Engineering, ICSE ’16. ACM, New York, pp 237–248. https://doi.org/10.1145/2884781.2884810
Matyunina J (2017) How do i apply machine learning in an android app? https://www.quora.com/How-do-I-apply-machine-learning-in-an-android-app (Last accessed: May 2018)
Minka TP (2007) A comparison of numerical optimizers for logistic regression. Unpublished paper available at http://research.microsoft.com/en-us/um/people/minka/papers/logreg/minka-logreg.pdf
Mizutani E, Dreyfus SE (2001) On complexity analysis of supervised mlp-learning for algorithmic comparisons. In: Neural networks. IEEE, vol 1, pp 347–352
OpenCV Team (2018) Android - opencv library. https://opencv.org/platforms/android/ (Last accessed: May 2018)
Padraig C, Sarah JD (2007) k-nearest neighbour classifiers. Technical Report UCD-CSI-2007-4, University College Dublin. https://csiweb.ucd.ie/files/UCD-CSI-2007-4.pdf
Pang C, Hindle A, Adams B, Hassan AE (2015) What do programmers know about the energy consumption of software? IEEE Software, pp 83–89. http://softwareprocess.ca/pubs/pang2015IEEESoftware.pdf
Pathak A, Hu YC, Zhang M (2011a) Bootstrapping energy debugging on smartphones: a first look at energy bugs in mobile devices. In: Proceedings of the 10th ACM Workshop on Hot Topics in Networks, HotNets-X, pp 5:1–5:6
Pathak A, Hu YC, Zhang M, Bahl P, Wang YM (2011b) Fine-grained Power Modeling for Smartphones Using System Call Tracing. In: Eurosys ’11. Salzburg, pp 153–168. https://doi.org/10.1145/1966445.1966460
Pereira R, Couto M, Saraiva Ja, Cunha J, Fernandes J a P (2016) The influence of the java collection framework on overall energy consumption. In: Proceedings of the 5th International Workshop on Green and Sustainable Software, GREENS ’16, pp 15–21
Pinto G, Castor F, Liu YD (2014) Mining Questions About Software Energy Consumption. In: MSR 2014, pp 22–31. https://doi.org/10.1145/2597073.2597110
Rasmussen K, Wilson A, Hindle A (2014) Green mining: energy consumption of advertisement blocking methods. In: Proceedings of the 3rd International Workshop on Green and Sustainable Software (GREENS 2014), pp 38–45. http://softwareprocess.ca/pubs/rasmussen2014GREENS-adblock.pdf
Release v2.1-20171001-lanchon . dexpatcher/dex2jar . github (2018) https://github.com/DexPatcher/dex2jar/releases/tag/v2.1-20171001-lanchon. (Last accessed: May 2018)
Saborido R, Beltrame G, Khomh F, Alba E, Antoniol G (2016) Optimizing user experience in choosing android applications. In: 2016 IEEE 23Rd international conference on software analysis, evolution, and reengineering (SANER), vol 1, pp 438–448. https://doi.org/10.1109/SANER.2016.64
Sahin C, Tornquist P, Mckenna R, Pearson Z, Clause J (2014) How does code obfuscation impact energy usage?. In: ICSME. IEEE Computer Society, pp 131–140
Sahin C, Pollock L, Clause J (2016a) From benchmarks to real apps: Exploring the energy impacts of performance-directed changes. J Syst Softw 117:307–316
Sahin C, Wan M, Tornquist P, McKenna R, Pearson Z, Halfond WG, Clause J (2016b) How does code obfuscation impact energy usage? J Softw Evol Process 28(7):565–588
Selby JWA (2011) Unconventional applications of compiler analysis. Ph.D. thesis, University of Waterloo
Sevarac Z, Goloskokovic I, Tait J, Carter-Greaves L, Morgan A, Steinhauer V (2016) Neuroph: Java neural network framework. http://neuroph.sourceforge.net/
Keerthi SS, Shevade SK, Bhattacharyya C, Murthy KRK (2001) Improvements to platt’s smo algorithm for svm classifier design. Neural Comput 13(3):637–649
Su J, Zhang H (2006) A fast decision tree learning algorithm. In: American association for artificial intelligence, vol 6, pp 500–505
TensorFlow (2016) Mobile tensorflow. https://www.tensorflow.org/mobile.html
Tomita TM, Maggioni M, Vogelstein JT (2015) Randomer forests. arXiv:1506.03410
Triposo (2016) Triposo. https://www.triposo.com/
Wan M, Jin Y, Li D, Gui J, Mahajan S, Halfond WG (2017) Detecting display energy hotspots in android apps. Software Testing Verification and Reliability 27(6). https://onlinelibrary.wiley.com/doi/abs/10.1002/stvr.1635
Webservices A (2018) Amazon aws machine learning https://aws.amazon.com/machine-learning/ (Last accessed: May 2018)
Weotta (2016) About weotta. http://www.weotta.com/about
Witten IH, Frank E (2011) Data Mining: Practical machine learning tools and techniques, 3rd edn. Morgan Kaufmann, Burlington
Woollaston V (2015) Customers really want better battery life. http://www.dailymail.co.uk/sciencetech/article-2715860/Mobile-phone-customers-really-want-better-battery-life-waterproof-screens-poll-reveals.html (last accessed: May 2018)
Yang Y, Zhang J, Kisiel B (2003) A scalability analysis of classifiers in text categorization. In: Proceedings of the 26th annual international ACM SIGIR conference on Research and development in informaion retrieval. ACM, pp 96–103
Acknowledgments
Andrea McIntosh was supported by a NSERC Undergraduate Student Research Award. Abram Hindle was supported by a NSERC Discovery Grant. Safwat Hassan was supported by Ahmed Hassan. We would like to thank Ahmed Hassan for his guidance.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by: Massimiliano Di Penta
Rights and permissions
About this article
Cite this article
McIntosh, A., Hassan, S. & Hindle, A. What can Android mobile app developers do about the energy consumption of machine learning?. Empir Software Eng 24, 562–601 (2019). https://doi.org/10.1007/s10664-018-9629-2
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10664-018-9629-2