Artificial Intelligence Applications and Innovations

Ilias Maglogiannis, Lazaros Iliadis, John MacIntyre, Manuel Domí­nguezIFIP 2023 · DOI · OpenAlex

The 19th Artificial Intelligence Applications and Innovations (AIAI 2023) conference proceedings provide a comprehensive overview of the current state of applied AI. This collection of 92 peer-reviewed papers, selected from 185 submissions, bridges the gap between cutting-edge theory and high-stakes practical deployment across diverse sectors such as healthcare, finance, cybersecurity, and sustainable energy.

Addressing the Translation Gap in AI

The primary focus of AIAI 2023 is the "translation" of AI from laboratory settings to real-world environments. While deep learning has achieved remarkable accuracy, its adoption in high-stakes fields like aviation or medicine is often hindered by three critical gaps:

  1. Transparency: The "black box" nature of state-of-the-art models makes it difficult for human experts to trust or verify decisions.
  2. Robustness: Models remain vulnerable to adversarial attacks and data poisoning, particularly in sensitive domains like medical imaging.
  3. Efficiency: The escalating computational and energy costs of training complex models are becoming increasingly unsustainable.

Methodological Landscape

To address these hurdles, the proceedings showcase novel algorithms prioritizing explainability, security, and computational efficiency. Key methodological highlights include:

  • Evolutionary Neural Architecture Search (NAS): Techniques for automatically designing deep learning architectures that are privacy-preserving and robust.
  • Interpretable-By-Design Models: A shift toward "prototype-based" deep learning that remains inherently understandable to humans, rather than relying on post-hoc explanations.
  • Visual Analytics (VA): Integrating machine learning with human reasoning through interactive data visualization.
  • Hybrid Architectures: Combining standard CNNs and Transformers with specialized techniques like Spiking Neural Networks (SNNs) for low-power edge computing.
A selection of AI algorithms and research areas discussed at the AIAI 2023 conference.

Domain-Specific Applications and Datasets

The research presented utilizes highly diverse and domain-specific datasets to solve complex problems:

  • Medical: MRI/CT scans for tumor segmentation and sensor data for Parkinson’s monitoring.
  • Social/NLP: Analysis of misinformation on social media and cyberbullying detection in multiple languages.
  • Industrial/Environmental: Hyperspectral imaging for recycling and wind turbine signals for fault diagnosis.
  • Financial: Forex and stock market data for reinforcement learning-based trading.

Key Results and Global Collaboration

AIAI 2023 maintained a rigorous peer-review standard, with a 40.5% acceptance rate for full papers. A defining characteristic of this year's proceedings is the global nature of the research effort, featuring contributions from 32 different countries across five continents, signaling a unified effort toward AI safety and practical application.

List of participating countries contributing to the AIAI 2023 proceedings, highlighting the global scale of AI research.

Future Directions and Limitations

Despite significant progress, several challenges remain. Key areas for future research include the development of legislative frameworks for autonomous AI, improving Zero-Shot Learning capabilities, and enhancing model performance in "non-stationary" real-world environments where data distributions shift over time. The ultimate goal is the development of "Autonomous Developmental AI"—systems capable of evolving their internal structures to recognize new data classes in real-time with minimal human intervention.

Summary updated

BibTeX
@inproceedings{maglogiannis2023artificial,
  author = {Ilias Maglogiannis and Lazaros Iliadis and John MacIntyre and Manuel Domí­nguez},
  title = {{Artificial Intelligence Applications and Innovations}},
  booktitle = {IFIP},
  year = {2023},
  doi = {10.1007/978-3-031-34111-3},
}
Figure