Journal of Big Data is calling for submissions to our Collection on Customization and fine-tuning of machine learning models. The special issue seeks papers on topics related to machine learning applications to sequential and temporal data, and other real-world applications, such as energy, manufacturing, traffic or finance. Novel applications of existing methods, such as reinforcement learning or transformers, are welcome. In the domain of LLMs, efficiency improvements, such as comparisons of prompt-based learning, fine tuning, parameter-efficient fine-tuning (PEFT) and Retrieval-Augmented Generation (RAG) systems are welcome.
Topics of interest include:
● Novel applications to sequence-based data, such as time series or language or biological data
● Hybrid machine learning methods incorporating novel feature extraction approaches
● Explainable AI for temporal or sequential data
● Knowledge graph applications to text or biomedical domains, named entity recognition, relation extraction
● PEFT techniques like adapter modules and knowledge distillation with applications to NLP tasks (question answering, summarization, sentiment analysis etc.)
● Comparisons between PEFT methods for applications requiring high accuracy, and prompt-based learning for low-resource settings requiring efficiency.
● Case studies where quantization for faster inference and fine-tuning have been successfully applied in practice
● Applications beyond NLP and text, such as computer vision tasks in robotics, finance, biomedical sciences, environmental sciences
Image credit: © [M] NicoElNino / Getty Images / iStock
