What can Android mobile app developers do about the energy consumption of machine learning?

McIntosh, Andrea K.; Hassan, Safwat; Hindle, Abram

doi:10.1007/s10664-018-9629-2

Cited by 46 publications

(24 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…(3) Retrofitting ML to edge devices:"I wanted to do advanced image processing on lowperformance, low-power, non-GPU equipped edge devices, " said developer S94. Retrofitting ML to applications on mobile devices is challenging due to limited processing power and limited battery life [95]. Quantization is a possible software solution to increase the performance of ML algorithms on such devices.…”

Section: Challenges When Retrofitting ML Librariesmentioning

confidence: 99%

See 1 more Smart Citation

Understanding Software-2.0

Dilhara

Ketkar

Dig

2021

ACM Trans. Softw. Eng. Methodol.

View full text Add to dashboard Cite

Enabled by a rich ecosystem of Machine Learning (ML) libraries, programming using learned models , i.e., Software-2.0 , has gained substantial adoption. However, we do not know what challenges developers encounter when they use ML libraries. With this knowledge gap, researchers miss opportunities to contribute to new research directions, tool builders do not invest resources where automation is most needed, library designers cannot make informed decisions when releasing ML library versions, and developers fail to use common practices when using ML libraries. We present the first large-scale quantitative and qualitative empirical study to shed light on how developers in Software-2.0 use ML libraries, and how this evolution affects their code. Particularly, using static analysis we perform a longitudinal study of 3,340 top-rated open-source projects with 46,110 contributors. To further understand the challenges of ML library evolution, we survey 109 developers who introduce and evolve ML libraries. Using this rich dataset we reveal several novel findings. Among others, we found an increasing trend of using ML libraries: The ratio of new Python projects that use ML libraries increased from 2% in 2013 to 50% in 2018. We identify several usage patterns including the following: (i) 36% of the projects use multiple ML libraries to implement various stages of the ML workflows, (ii) developers update ML libraries more often than the traditional libraries , (iii) strict upgrades are the most popular for ML libraries among other update kinds, (iv) ML library updates often result in cascading library updates, and (v) ML libraries are often downgraded (22.04% of cases). We also observed unique challenges when evolving and maintaining Software-2.0 such as (i) binary incompatibility of trained ML models and (ii) benchmarking ML models. Finally, we present actionable implications of our findings for researchers, tool builders, developers, educators, library vendors, and hardware vendors.

show abstract

Section: Challenges When Retrofitting ML Librariesmentioning

confidence: 99%

“…Further, they observed that contributions from both academics and companies fuel the ML ecosystem. Mcintosh et al [95] studied the power consumption of ML algorithms on mobile devices and revealed that many ML algorithm implementations consume more power to train than to evaluate.…”

Section: Studies On Software-20mentioning

confidence: 99%

Understanding Software-2.0

Dilhara

Ketkar

Dig

2021

ACM Trans. Softw. Eng. Methodol.

View full text Add to dashboard Cite

show abstract

“… (K)nowledgable communities: skilled Android developers report how they have dealt with excessive battery depletion and how they have solved them, whether in blog posts or on popular exchange platforms such as StackOverflow or Google Groups.  (R)esearch literature: studies regularly focus on very specific aspects in order to identify hotspots, like cryptography [6], advertising [7], logging [8], network traffic [9], display [10], ML [11] etc.…”

Section: A Green Bugs Overviewmentioning

confidence: 99%

Enforcing green code with Android lint

Goaer

2020

Proceedings of the 35th IEEE/ACM International Conference on Automated Software Engineering Workshops

View full text Add to dashboard Cite

Nowadays, energy efficiency is recognized as a core quality attribute of applications (apps) running on Androidpowered devices constrained by their battery. Indeed, energy hogging apps are a liability to both the end-user and software developer. Yet, there are very few tools available to help developers increase the quality of their native code by ridding it of energy-related bugs. Android Studio is the official IDE for millions of developers worldwide and there's no better place to enforce green coding rules in everyday projects. Indeed, Android Studio provides a code scanning tool called Android lint that can be extended with lacking green checks in order to foster the design of more eco-responsible apps.

show abstract

“…e energy consumption in mobile devices has been studied in different works. e experimental studies cope with energy consumption in different contexts and with different purposes [12,27,[29][30][31][39][40][41][42][43][44][45][46][47][48][49][50][51]. ese studies are conducted to help developers about being concerned of energy consumption and contribute to provide alternative choices more energy saving.…”

Section: Related Workmentioning

confidence: 99%

An Energy Efficiency Study of Web-Based Communication in Android Phones

Ayala

Amor

Fuentes

2019

Scientific Programming

View full text Add to dashboard Cite

Currently, mobile devices are the most popular pervasive computing devices, and they are becoming the primary way for accessing Internet. Battery is a critical resource in such personal computing gadgets, network communications being one of the primary energy consuming activities in any mobile app. Indeed, as web-based communication is the most used explicitly or implicitly by mobile devices, HTTP-based traffic is the most power demanding one. So, mobile web developers should be aware of how much energy demands the different web-based communication alternatives. e goal of this paper is to measure and compare the energy consumption of three asynchronous HTTP-based methods in mobile devices in different browsers. Our experiments focus on three HTTP-based asynchronous communication models that allow a web server to push data to a client browser through a HTTP/1.1 interaction: Polling, Long Polling, and WebSockets. e resulted measurements are then analysed to get more accurate understanding of the impact of the selected method, and the mobile browser, in the energy consumption of the asynchronous HTTP-based communication. e utility of these experiments is to show developers what are the factors and settings that mostly influence the energy consumption when different web-based asynchronous communication methods are used, helping them to choose the most beneficial solution if possible. With this information, mobile web developers should be able to reduce the power consumption of the front-end of web applications for mobile devices, just selecting and configuring the best asynchronous method or mobile browser, improving the performance of HTTP-based communication in terms of energy demand.

show abstract

What can Android mobile app developers do about the energy consumption of machine learning?

Cited by 46 publications

References 40 publications

Understanding Software-2.0

Understanding Software-2.0

Enforcing green code with Android lint

An Energy Efficiency Study of Web-Based Communication in Android Phones

Contact Info

Product

Resources

About