Computer Science - Conference Items

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 5 of 388
  • Item
    UNCCER: unified network for cancer classification and efficient representation using microarray data
    (Institute of Electrical and Electronics Engineers (IEEE), 2024-12-26) Younis, Haseeb; Brahmi, Imane; Byrne, Jonathan; Minghim, Rosane; Science Foundation Ireland
    In recent years, there has been a significant advance in the use of machine learning (ML) techniques to extract gene expression data from microarray databases, particularly in cancer-related research. There no unified method for classifying cancer microarray data, even after ML adoption. Due to the high dimensionality of microarray data, it is difficult to extract the relevant features and provide insights that can be helpful in identifying cancer types and stages. In this paper, we propose a Unified Network for Cancer Classification and Efficient Representation (UNCCER) using Deep Learning (DL) on cancer microarray data. To implement this methodology, we employed a microarray database (CuMiDa) that has 78 carefully curated datasets for different types of cancers. Our single model has the capability to learn the patterns, cluster instances into their corresponding classes, and classify the cancer. We also used the Uniform Manifold Approximation and Projection (UMAP) to visualise, in low dimension, the instance separation both on original data and transformed data by our methodology. Using the proposed methodology, we achieved average 94% average accuracy, precision, recall, F1 Score, and 91% G-Mean.
  • Item
    MANet: a deep learning framework for multi-cancer microarray analysis, classification, and visualization
    (Institute of Electrical and Electronics Engineers (IEEE), 2024) Younis, Haseeb; Azeem, Muhammad; Ronan, Isabel; Minghim, Rosane
    Machine learning (ML) methods have been used much more frequently in recent years to extract gene expression data from microarray studies, especially in cancer research. Even after the continued interest in applying ML to scientific cancer research, there is still no universal approach for categorizing cancer microarray data. A system is needed that can detect and classify a normal profile and a cancer profile, specifying the type of cancer. Due to the variance and high dimensionality of microarray data, it is difficult to extract the relevant descriptors and provide insights that can be helpful in identifying cancer types and stages. In this paper, we proposed MANet: a methodology using cancer microarray data based on Deep Learning (DL) to classify 13 different types of cancers as well as normal profiles. To implement this methodology, we used a Curated Microarray Database (CuMiDa) that has 78 datasets for different types of cancers. Due to the diverse feature vectors for each dataset, we used Principal Component Analysis (PCA) for uniform feature engineering. Our single model has the capability to learn the patterns, cluster instances into their corresponding classes and classify the cancer. We also used the Uniform Manifold Approximation and Projection (UMAP) to visualise the instance separation on original data. Additionally, this UMAP visualises segregation done by our methodology in low dimensions. Using the proposed methodology, we achieved approximately 80% average accuracy, precision, recall, and F1 Score for 14 classes using a single model
  • Item
    Fine-tuning generative pre-trained transformers for clinical dialogue summarization
    (Institute of Electrical and Electronics Engineers (IEEE), 2025-01-17) Ronan, Isabel; Tabirca, Sabin; Science Foundation Ireland
    Automated clinical dialogue summarization can help make health professional workflows more efficient. With the advent of large language models, machine learning can be used to provide accurate and efficient summarization tools. Generative Pre-Trained Transformers (GPT) have shown huge promise in this area. While larger GPT models, such as GPT-4, have been used, these models pose their own problems in terms of precision and expense. Fine-tuning smaller models can lead to more accurate results with less computational expense. In this paper, we fine-tune a GPT-3.5 model to summarize clinical dialogue. We use both default hyperparameters along with manual hyperparameters for comparison purposes. We also compare our default model to past work using ROUGE-1, ROUGE-2, ROUGE-L, and BERTScores. We find our model outperforms GPT-4 across all measures. As our fine-tuning process is based on the smaller GPT-3.5 model, we show that fine-tuning leads to more accurate and less expensive results. Informal human observation also reveals our notes to be of acceptable quality.
  • Item
    “ImmediateShortTerm3MthsAfterThatLOL”: Developer secure-coding sentiment, practice and culture in organisations
    (2025-05) Ryan, Ita; Roedig, Utz; Stol, Klaas-Jan; Science Foundation Ireland
    As almost all areas of human endeavour undergo rapid digital transformation, secure coding is increasingly important to personal, commercial and national security. Yet studies have shown that software developers do not always prioritise or even understand security. Our large survey of organically sourced coders (n=863) examines how software developers currently experience secure coding in the workplace. We found that developers express an interest in secure coding, display basic security knowledge, and turn to their managers and teams first for help with security concerns. We found that developer secure coding sentiment and security practice do not correlate with organisational statistics such as size, but do correlate weakly with measures of security culture, indicating that organisational security support goes hand-in-hand with secure development. Most developers would look for help in-house if they had security concerns. Investigating the effects of code breaches, we found that for almost half of cases, code security does not increase, or increases only for a short time.
  • Item
    Evidence theory-based trust management for the Social Internet of Vehicles
    (Institute of Electrical and Electronics Engineers (IEEE), 2024-09-09) Shamaeian, Nasrin; Pesch, Dirk; Science Foundation Ireland
    The Social Internet of Vehicles (SIoV) is a concept combining the principles of vehicular and social networks, where entities, such as vehicles, drivers, passengers and infrastructure, share information not only for intelligent transportation or cooperative mobility needs, but also using social network principles. Trust in the information exchanged between vehicles in a vehicular network is paramount to achieving safety and reliability of transportation. We propose a trust management model for SIoV that integrates entity trust from direct interactions between vehicles, indirect trust from recommendations, and social trust reflecting the drivers’ social attributes. We utilize Dempster–Shafer Theory to effectively manage inherent uncertainties within this network, enabling robust aggregation of various trust evidences. Our simulation results show the effectiveness of our model in accurately identifying and mitigating malicious entities within the network performing trust-related attacks. Published in: 2024 IEEE 49th Conference on Local Computer Networks (LCN)