01. Kandidaatin tutkinnon opinnäytteet / Bachelor’s theses
Permanent URI for this communityhttps://aaltodoc.aalto.fi/handle/123456789/1
Browse
Browsing 01. Kandidaatin tutkinnon opinnäytteet / Bachelor’s theses by Degree programme/Major subject "Aalto Bachelor’s Programme in Science and Technology"
Now showing 1 - 20 of 200
- Results Per Page
- Sort Options
- Addressing Challenges in Reinforcement Learning With Model-Based Methods
Perustieteiden korkeakoulu | Bachelor's thesis(2023-05-07) Karhula, Sandro - Addressing Risks in Artificial Intelligence Act and in Future Quantum Computing Legislation
Perustieteiden korkeakoulu | Bachelor's thesis(2024-09-06) Korpimaki, IlariArtificial intelligence (AI) and quantum computing (QC) are two trans-formative technologies with the potential to reshape global public policies. Although distinct, both face similar regulatory challenges, particularly in data protection, national stability, and market power. Major corporations and geopolitical competition drive their development, making effective regulation essential. In response to this need, this thesis explores the intersection of artificial intelligence and quantum computing, focusing on their similarities and associated risks. The research aims to determine how the EU Artificial Intelligence Act (AIA) addresses various risks and to apply these insights to future EU regulation of quantum computing. To achieve this, a state-of-the-art literature review was conducted, examining the current development and potential challenges of both technologies. The findings aim to provide a foundation for developing effective quantum computing regulations. - Adversarial robustness of GPT-3.5 Turbo
Perustieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Nguyen, AnhThe rapid advancement of artificial intelligence (AI) models has brought forth a critical need for a thorough examination of potential ethical and security concerns. However, many ethical issues regarding AI are being overlooked, including misinformation, bias, and accuracy. Within the scope of robustness, the study aims to assess the consistency of GPT’s output given the diversity of inputs. The primary objective is to construct a comprehensive framework for assessing the robustness of large language models (LLMs), with a specific emphasis on GPT 3.5. The research takes a proactive stance by developing experiments and methods using Python coding language, incorporating literature review and data analysis. This study answered the question how the algorithmic robustness of Large Language Models can, particularly GPT 3.5 turbo , be effectively assessed. By delving deeply into the assessment of robustness, the research seeks to address challenges associated with the widespread use of powerful language models, fostering a more secure, ethical, and transparent landscape in the field of artificial intelligence. - Analysing Customer Feedback using Sentiment Analysis for Digital Service Development
Perustieteiden korkeakoulu | Bachelor's thesis(2022-12-16) Niva, Verna - An Analysis of Boundary Conditions in Partial Differential Equations using Finite Difference Method
Insinööritieteiden korkeakoulu | Bachelor's thesis(2021-09-06) Lasinski, Dominik - Analysis of dedicated techniques for source code embedding
Perustieteiden korkeakoulu | Bachelor's thesis(2022-04-15) Strozanski, Pawel - Analysis of Parameters Affecting the Operation of a Switched Reluctance Motor
Insinööritieteiden korkeakoulu | Bachelor's thesis(2024-09-06) Svanidze, NikolozThis thesis studies the relationship of torque generated by a static switched reluctance motor with the input current, number of windings, relative permeability and air gap of the said motor. It presents findings based on the finite element method. The thesis finds and presents quadratic dependencies for input current and number of windings. Patterns in air gap and relative permeability graphs are noted and explained. The results align with the work published by other researchers. - Application of Machine Learning in Product Structure for Mass Customisation: Model Structure Validation with Large Language Model
Insinööritieteiden korkeakoulu | Bachelor's thesis(2024-12-27) Nguyen, HuyenThis thesis investigates the application of machine learning, specifically fine-tuned large language models (LLMs), in validating product structures for mass customisation. The research focuses on kitchen ventilation systems, employing GPT-4o to streamline product structure validation. A dataset of 250 samples, including correct and misconfigured examples, was used to train the model. The study evaluates the impact of prompt engineering, dataset size, and fine-tuning strategies, highlighting the challenges of limited data and input constraints. Results indicate that the fine-tuned model effectively identifies misconfigurations, provides actionable recommendations, and understands the principles of mass customisation. The findings demonstrate the potential of LLMs in enhancing knowledge transfer and complexity management in product configuration systems. While limitations such as dataset size and model scalability remain, this research establishes a foundation for extending machine learning solutions to broader manufacturing contexts. - Application of Natural Language Processing in Financial News Sentiment Analysis for Stock Price Prediction
Perustieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Nguyen, HiepThis thesis studies the application of Natural Language Processing (NLP) in the analysis of financial news sentiment and its subsequent impact on stock price prediction. With the increasing complexity of the financial market, the need for advanced computational techniques to predict stock price movement is evident. This study systematically reviews current research findings on the efficacy of NLP methods in analyzing the sentiment of financial news and stock price movements. By conducting a systematic literature review on the research from the past six years, this review emphasizes the development of sentiment analysis and NLP techniques and evaluates their predictive power. A total number of 33 papers were chosen for this review. Key findings suggest that due to the recent advancement, particularly the introduction of the transformer model, the focus of NLP in stock prediction has shifted from traditional statistical-based feature representation to learning-based embedding methods. The conclusion addresses the potential of sentiment analysis as a predictive tool and suggests directions for future research. This emphasizes the need for innovative NLP applications in the financial domain to enhance investment strategies and market understanding. - The Application of Reinforcement Learning (RL) in Autonomous Ship and Collision Avoidance: A Systematic Literature Review
Insinööritieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Rezaei, ZainIn recent years, research on the application of reinforcement learning (RL) to improve the intelligence of ships has increased significantly. This study presents a systematic literature review that examines the applications of RL in the field of autonomous ships and collision avoidance. In particular, it answers the research questions of what different RL models are used, what different inputs they process and what range of outputs they produce. In addition, the simulation environments and the choice of hyperparameters that researchers use to train these models are studied. Finally, this study highlights the current implementation challenges in this area to ensure the safe and robust use of RL models in autonomous vessels. - Applications of Causality in Reinforcement Learning
Perustieteiden korkeakoulu | Bachelor's thesis(2021-09-10) Ray, Atreya - Applying machine learning to metabolite identification: Comparing Robust Loss Functions
Perustieteiden korkeakoulu | Bachelor's thesis(2022-05-06) Pham, Binh - Artificial Intelligence and Game Theory in No-Limit Texas Hold’em Poker
Perustieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Heidmets, MarkAs with many board and card games, researchers have attempted to solve poker using both game theory and artificial intelligence (AI). However, poker is an incomplete information game that is often played with over two players, meaning that solving the game becomes more complex than for complete information heads-up (two-player) games, such as chess. For several decades, efforts to develop a superhuman AI in multiplayer poker were unsuccessful, until recently when first such AI was created. Despite that, multiplayer poker is still not completely solved, and many computational and game-theoretical challenges persist. The no-limit Texas hold’em is a game-theoretically complex and captivating version of poker, including computationally challenging aspects such as imperfect information and multiplayer dynamics. Additionally, no-limit hold’em is currently the most popular version of poker, thus receiving more focus academically as well. The thesis studies AI and game theory in no-limit Texas hold’em poker as well as the computational challenges associated with the AI models by conducting a literature review of the relevant sources in the field. The paper covers the theoretical aspects of game theory in relation to poker as well as the technological advancements in developing a competitive poker model. The algorithm used in current state-of-the-art models, counterfactual regret minimization, is discussed in more detail. Due to the computational difficul- ties with no-limit hold’em game tree, several optimizations, such as abstractions and improved versions of counterfactual regret minimization must be utilized to create a capable AI for no-limit hold’em poker. - Artificial Intelligence for Prostate Cancer Screening and Diagnosis: A State-of-the-art Review
Perustieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Do, ChauProstate cancer (PCa) is one of the most prevalent forms of cancer affecting men worldwide. This thesis conducts a state-of-the-art review of AI-based approaches for PCa screening and diagnosis, focusing on three major types of input data: prostate-specific antigen (PSA) screening data, magnetic resonance images (MRIs), and histopathology images. The purpose of this thesis is to determine the current extent of research into AI for PCa screening and diagnosis, highlighting recent advancements, future prospects, and remaining challenges. The selected articles show wide variability in study designs, data sources, model architectures, and evaluation approaches. Key findings suggest that even though recent research mostly focuses on AI for medical image analysis, PSA data still carries valuable information and can be combined with MRI data and other clinical variables to improve diagnostic performance. AI models inputting MRIs have demonstrated diagnostic performance surpassing the PI-RADS (prostate imaging-reporting and data system) on several coarse-level PCa classification tasks. Similarly, AI models employed for Gleason grading of histopathology images have been reported to match pathologists' performance and to improve inter-rater agreement. In addition to the literature review, an analysis consisting of 11 articles was conducted to identify the relationship between the AUC values, cohort sizes, and model types of models differentiating between clinically significant PCa (csPCa) and non-csPCa using MRIs. The analysis reveals a statistically significant positive correlation (Kendall's tau = 0.673, 95% CI: 0.348-0.944, p = 0.020) between the AUCs and the cohort sizes, suggesting that increasing cohort size might have a positive impact on performance. In addition, permutation tests with the t-test statistic indicate a statistically significant difference (t = 2.297, p = 0.032) between the mean AUC values of traditional machine learning models and deep learning models, suggesting that deep learning models might be capable of achieving better performance than traditional machine learning models on such medical image analysis tasks. Despite the great potential and significant progress of research into AI for PCa, several challenges remain, including the shortage of data, label noise, and wide variability in study implementation and evaluation. These problems call for the application of novel machine-learning techniques and collaborative research endeavors. - Assessing the Effectiveness of Biological Processes for Micropollutant Removal in Wastewater Treatment
Perustieteiden korkeakoulu | Bachelor's thesis(2023-05-19) Nyman, Jesse - Asymptotic curves on minimal surfaces
Insinööritieteiden korkeakoulu | Bachelor's thesis(2022-04-15) Nagyvaradi, Balazs - Automatic Code Review with Deep Learning
Insinööritieteiden korkeakoulu | Bachelor's thesis(2022-04-15) Pajula, Ilari - Automatic Hardware-aware Optimization of Fault-tolerant Quantum Circuits
Perustieteiden korkeakoulu | Bachelor's thesis(2024-04-26) Do, HuyenThis thesis explores the calibration of quantum error correction codes (QECC) by focusing on optimizing plaquette circuits at the native gate level and assessing their error rates on real quantum hardware. The research underscores the importance of native gate optimization to boost the performance and reliability of quantum computations. A crucial part of the study involves experiments on the repetition of plaquette circuits using two ancilla qubit configurations: Type A, involving fresh ancilla qubits for each repetition, and Type B, reusing ancilla qubits. Based on the data collected, an error model was formulated. A significant result is that the error rate for a single plaquette operation using IonQ’s Aria1 Quantum Processing Unit (QPU) was 1.5%, surpassing the essential 1% threshold for effective quantum error correction, thus indicating a need for further improvements. The thesis also demonstrates the variability in QECC implementation across different quantum architectures, emphasizing the challenges in establishing universally applicable error rates. The study proposes to extend this research to evaluate the error rates of plaquette circuits on other types of quantum hardware, such as superconducting quantum computers. Future work will develop a broader understanding of QECC’s effective implementation across various quantum computing platforms, enhancing the robustness and scalability of quantum error correction methods. - Bayesian Regression Techniques for High-Dimensional Financial Time Series Data Structures
Perustieteiden korkeakoulu | Bachelor's thesis(2024-12-13) Vintola, OttoAs the world adpots a culture of data-driven decision-making, the number of high-dimensional datasets increases. However, high dimensionality might bring issues such as ill-posedness, conflated models, and overfitting, thus necessitating shrinkage methods to reduce the dimensionality by selecting or penalizing the utilized features. Analogously, this thesis aims to find the most impactful stocks in a high-dimensional portfolio. Previous research regarding this topic has explored portfolios, high-dimensionality, sparsity, and Bayesian methods. Nevertheless, the literature concentrating on Bayesian shrinkage for high-dimensional portfolio data, is limited. Hence, this thesis aims to uncover sparsity present in one of the most popular portfolios in the world, the S&P500. Moreover, the training dataset consists of daily observations regarding the S&P500 and its constituent stocks across the years [2018, 2022], while the validation set is for the year 2023 alone. As the chosen methodology two common shrinkage priors, horseshoe and spike-and-slab, are placed on the Bayesian regression model. Conducting the trials reveals, that spike-and-slab provides superior predictive power over horseshoe. Spike-and-slab requires 174 unique stock ticers, corresponding to 180 regressors, for adequate predictive power measured by adjusted coefficient of determination. The number of tickers could be researched through the selections made by credible intervals, however, they provide a lower bound for the number non-zero regressors required. Consequently, the contribution of this thesis is the uncovered sparsity in the S&P500 with Bayesian methods, along with a suggestion for the shrunk variable selection method by comparing the adjusted coefficient of determination and credible intervals. - Bayesian reinforcement learning: A comparative study on Bayesian model-free and model-based approaches
Perustieteiden korkeakoulu | Bachelor's thesis(2020-12-18) Lebedeff, Anna