Sirvi Autor "Anwar, Hina, juhendaja" järgi

Nüüd näidatakse 1 - 7 7

Benchmarking Energy and Performance of Standard Machine Learning Libraries: An Empirical Study
(Tartu Ülikool, 2025) Kont, Kadri-Ketter; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Programmeerimiskeele ja teegi valiku mõju masinõppeülesannete energiakulule ei ole seni põhjalikult uuritud. Lõputöö eesmärk oli võrrelda kolme masinõppes levinud programmeerimiskeelt ja nende teeke, keskendudes energiatõhususele, käitusajale ja mudeli täpsusele. Klassifitseerimisülesande andmestiku põhjal rakendati standardteekide abil igas keeles viis masinõppealgoritmi. Uurimuses analüüsiti, kuidas keele ja teegi valik mõjutab energiatarbimist, käitusaega ja täpsust, ning uuriti nende näitajate vahelisi kompromisse. Tulemuste kinnitamiseks viidi läbi statistiline analüüs, mille abil võrreldi algoritmide energiakasutust ja jõudlust erinevate implementatsioonide puhul. Töö tulemusena tehti ülevaade, kuidas standardteekide valik mõjutab masinõppe algoritmide energiatõhusust.
Code Smarter Not Harder: Measuring Energy Efficiency of LLM-Generated Code
(Tartu Ülikool, 2025) Süleymanova, Fidan; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
As large language models (LLMs) become more integrated into software development, questions arise not only about correctness but also about the energy behavior of the code they generate. This thesis investigates how LLM-generated code compares to human-written solutions in terms of runtime and energy efficiency, and if minimal prompting can improve the energy profile of the output. Using four real-world programming tasks which is implemented in Python, JavaScript, and Rust, we benchmarked code produced by three LLMs, GPT-4o, LLaMA-3.3-Instruct, and Qwen 2.5-Coder against curated human-written submissions. Energy and runtime were measured using low-level CPU profiling tools in a tightly controlled environment. Results show that LLMs can often approach the performance of human-written code, but rarely exceeds it, with GPT-4o delivering the most consistent outcomes. Prompting for energy efficiency had a measurable but inconsistent effect, improving results in some cases while degrading them in others. Statistical analysis found no significant group-level differences between LLM-generated code and human-written code, but small task-level variations were also observed. These findings highlight both the potential and the current limits of prompt-based energy optimization and they provide a reproducible framework for future research in energy-aware code generation.
Emotion-oriented Game-based Fitness App for Diabetes Patients
(Tartu Ülikool, 2023) Shrestha, Monika; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Diabetes is a growing health concern globally, and managing the condition requires regular monitoring of glucose levels and physical activity. Conventional methods of monitoring and exercise often lack the motivation required to keep patients engaged and committed. This thesis proposes the development of an emotion-oriented game-based fitness application - DiaBeatIt, specifically designed for diabetes patients to support them in maintaining a healthy lifestyle despite their disease through the use of exercise such as running or walking. DiaBeatIt combines elements of gamification to provide a fun and interactive way of physical activity which includes walking on real-world maps and solving diabetes-related quizzes. In this research, the goal model is used to develop DiaBeatIt. The created application is subsequently tested on a variety of diabetes patients to see if the emotional goals of the patients are found to match the emotional goals obtained before and after the application trial.
Energy Matters: Evaluating JavaScript Asynchronous Patterns for Green Development
(Tartu Ülikool, 2025) Kasenõmm, Artur; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
IKT-sektori suurenev energiatarbimine koormab keskkonda ja majandust. Veebiarenduses domineeriv JavaScript ei kuulu teiste programmeerimiskeeltega võrreldes kõige energiasäästlikumate hulka. Ehkki tarkvara optimeerimisel pööratakse enamasti tähelepanu tehnoloogia valikule või algoritmide täiustamisele, on ühe ja sama funktsionaalsuse saavutamiseks kasutatavate programmeerimismustrite energiamõju ühe keele piires seni vähe käsitlust leidnud. Käesolevas lõputöös võrreldakse JavaScripti asünkroonsete mustrite callback, promise ning async/await energiakulu. Uurimus põhineb Node.js käituskeskkonnas ja Perfi mõõtmistööriistaga tehtud katsetel milles jäljendati HTTPpäringuid ja faili kirjutamist. Statistilise andmeanalüüsi käigus leidsime, et fikseeritud viitajaga HTTP-päringute korral tarbis callback muster vähem energiat (ligikaudu 6–7%) kui async/await lahendus. Failidesse kirjutamise puhul aga mustrite vahel märkimisväärset erinevust energiatarbimises ei täheldatud. Need tulemused viitavad sellele, et callback muster võib teatud olukordades olla energiatõhusam valik. Sellegipoolest tuleb arvestada promise ja async/await mustrite eelistega koodi loetavuses, hooldatavuses ning veakäsitluses. Enamasti kaaluvad need eelised üles väikese energiasäästu, mistõttu on promise ja async/await mustrid paljudes olukordades tõenäoliselt parem valik. Lõplik otsus sõltub siiski konkreetsest olukorrast ning sobiva tasakaalu leidmisest mõningase energiasäästu ja arendaja töö mugavuse ning tõhususe vahel.
Lab package: Mobile application security testing
(Tartu Ülikool, 2022) Pank, Geitrud; Pfahl, Dietmar, juhendaja; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
This thesis aims to create lab materials for the University of Tartu's course "Software testing" (LTAT.05.006) about mobile application security testing. This thesis gives background information on the course, on security testing in general and specifically mobile application security testing, describes the materials and lab execution using the materials, states, analyses the student feedback form and lab supervisors' feedback, and suggests future improvements. The lab took place in the spring semester of 2022.
Pocket Trainer for Equestrians
(Tartu Ülikool, 2023) Sõer, Kati; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Equestrian sports are gaining more popularity, but there are not enough professional trainers to reach everybody. The problem of equestrians finding proper trainers or trainers at all is rising. As equestrians often train alone, they experience problems they do not know how to solve themselves. To solve the problem, Pocket Trainer, a mobile application in Estonian, was developed to assist equestrians with problems they experience during training. The user can read helpful information in Estonian about solving the most common problems, look up different jumping and pole work exercises, and calculate the appropriate distances for setting up exercises. A certified trainer evaluated the information included in the application and an experienced equestrian evaluated the application from a user’s perspective. Overall the tester was satisfied with the app and the information it offers.
The Hidden Cost of Autosave: Measuring Energy Usage in Text Editors
(Tartu Ülikool, 2025) Küüsvek, Maria; Anwar, Hina, juhendaja; Tartu Ülikool. Loodus- ja täppisteaduste valdkond; Tartu Ülikool. Arvutiteaduse instituut
Software energy consumption is an increasingly important but underexplored aspect of global energy usage. With the increasing usage of portable devices, ICT solutions are expected to become a significant contributor to global energy consumption. However, developers currently have limited guidance on how to mitigate this impact. Background processes in software are easy to overlook, and developers may unknowingly choose energy-inefficient implementations of such features. This thesis looks into one such process: Autosaving in the case of desktop text editors. An empirical investigation was carried out to explore the real-world impact of overlooked energy inefficiencies. Common autosave operations (such as basic write, change tracking, and logging) were extracted and implemented from three open source text editing applications written in Python (novelWriter, Mu, and Leo). Using the Linux tool Perf, 900 measurements were carried out across 30 test scenarios, tested under different file sizes (Empty, 5KB, 50KB). Statistical analysis revealed that certain implementations produce a considerable energy overhead. Specific built-in features and write methods increase energy usage marginally. However, the most substantial energy savings were observed in the case of reducing save frequency, with one application showing potential to reduce its energy consumption by 83%. Based on these insights, four practical guidelines were proposed to help developers reduce the environmental impact of their software. The results highlight how small design choices can yield meaningful improvements in energy efficiency, supporting the broader goals of green software development.