Symmetry/Asymmetry Studies in Data Mining & Machine Learning of Large Language Models

Dear Colleagues,

The landscape of data mining and machine learning has been dramatically reshaped by the advent of large language models (LLMs). These powerful, data-hungry models have pushed the boundaries of what is possible, achieving unprecedented performance in tasks ranging from natural language understanding to image generation. However, as we delve deeper into the workings of LLMs, the intricate interplay between symmetry and asymmetry becomes increasingly apparent, presenting both opportunities and challenges for future research. Traditionally, symmetry played a central role in data mining and machine learning, with algorithms often seeking to identify recurring patterns and regularities. This approach, while effective in certain domains, has limitations when dealing with the vast and complex datasets that LLMs consume. Asymmetry, in contrast, offers a more nuanced perspective, acknowledging the inherent variability and irregularity within real-world data.

This Special Issue focuses on exploring the implications of symmetry and asymmetry in the context of large language models:

• LLMs often rely on massive, diverse datasets that exhibit inherent asymmetry. How can we leverage this asymmetry to improve data representation and encoding within LLMs?
• Can we develop new data representation techniques that specifically capture asymmetrical relationships?
• How can we incorporate domain-specific knowledge to address asymmetries in the data?
• The design of LLMs inherently involves balancing symmetric and asymmetric elements, from the architecture of neural networks to the training process. How can we leverage the interplay between symmetry and asymmetry to optimize model performance and efficiency?
• Can we design new neural network architectures that are more adept at handling asymmetrical data?
• How can we utilize asymmetric training strategies to enhance model performance and robustness?
• Explainability and Interpretability: LLMs are often criticized for their lack of transparency. Can the principles of symmetry and asymmetry contribute to developing more explainable and interpretable LLMs, making their decisions more understandable?
• How can we use symmetry and asymmetry to identify key features and relationships that drive LLM predictions?
• Can we develop new visualization techniques that highlight the interplay between symmetric and asymmetric patterns in LLM decision-making?
• Bias and Fairness: The vast datasets used to train LLMs can contain inherent biases, which may manifest as asymmetrical patterns. How can we use our understanding of symmetry and asymmetry to mitigate bias and promote fairness in LLMs?
• How can we identify and mitigate asymmetrical biases that may be present in the training data?
• Can we design new methods to measure and quantify the impact of symmetry and asymmetry on fairness in LLM outputs?
• Beyond NLP and Vision: LLMs are increasingly being applied in diverse domains beyond natural language processing and computer vision. How do the concepts of symmetry and asymmetry manifest in these new applications, and how can they be leveraged to improve model performance?
• How can we apply the principles of symmetry and asymmetry to develop LLMs for tasks like time-series analysis, scientific data analysis, or drug discovery?

Dr. Shaolin Zhu
Dr. Lijie Wen
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Symmetry is an international peer-reviewed open access monthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• symmetry
• asymmetry
• data mining
• machine learning
• large language models
• deep learning