2020 IEEE CIS Awards [Society Briefs]

Cesare AlippiIEEE Comput. Intell. Mag 2020 · DOI

Abstract: The 2020 IEEE Computational Intelligence Society (CIS) Awards recognize the year’s most influential figures and research contributions across fuzzy systems, evolutionary computation, and neural networks. This article profiles the award-winning breakthroughs that are shifting artificial intelligence from static, handcrafted models toward adaptive, self-learning systems capable of navigating complex, real-world environments.

Introduction

The 2020 IEEE CIS Awards celebrate a diverse range of contributions, from foundational mathematical theories to practical applications in healthcare and game design. A central theme among this year's winners is the effort to overcome "brittleness" in AI—the tendency for traditional machine learning to fail when faced with sparse data or changing environments.

Pioneering Adaptive and Evolutionary Systems

A major highlight of the awards is the recognition of researchers moving toward biologically inspired, adaptive architectures. Laszlo T. Koczy, recipient of the Fuzzy Systems Pioneer Award, was honored for his foundational work in rule interpolation. This method addresses gaps in sparse fuzzy models, allowing systems to reason effectively even when specific rules are missing.

In the realm of evolutionary computation, Risto Miikkulainen received the Pioneer Award for his work in neuroevolution. Miikkulainen posits that evolution is the key to surpassing the current limits of deep learning. By using evolutionary algorithms to "grow" neural network architectures, his methods allow systems to optimize themselves for complex tasks like robotics and natural language processing through biologically plausible adaptation.

Risto Miikkulainen, recipient of the 2020 Evolutionary Computation Pioneer Award, recognized for his work in neuroevolution.

Next-Generation Reinforcement Learning and Game AI

The Outstanding Early Career Awards highlight researchers pushing the boundaries of how machines learn and interact. Zhen Ni was recognized for developing Goal Representation Adaptive Dynamic Programming (GrADP). This framework introduces a "reference network" to standard actor-critic architectures, enabling agents to adaptively update internal goals. This move toward internalized motivation is critical for robustness in unstructured environments where external rewards may be sparse.

Julian Togelius was honored for his innovative use of video games as a benchmark for general intelligence. Togelius argues that games are the ideal "Drosophila" (fruit fly) for AI research—faster and more complex than robotics, yet designed specifically to challenge human-level cognition. His work in search-based procedural content generation allows AI to design game levels, which in turn serve as rigorous tests for the generalization capabilities of reinforcement learning agents.

Julian Togelius, recipient of the 2020 Outstanding Early Career Award, for his contributions to game AI and procedural content generation.

Community Leadership and Service

The growth of computational intelligence relies not only on technical breakthroughs but also on the strength of the professional community. Pau-Choo (Julia) Chung received the Meritorious Service Award for her leadership in membership development and her efforts to increase the visibility of the society, particularly among young professionals. Her work ensures that the field remains vibrant and inclusive as it tackles new challenges in healthcare and logistics.

Key Results and Future Directions

The awards also highlight seminal literature, such as the massive benchmarking of Long Short-Term Memory (LSTM) variants and the widely used PlatEMO platform for multi-objective optimization. Collectively, the featured pioneers have published over 1,150 articles, demonstrating the academic depth of the field.

Looking forward, the society identifies several critical hurdles:

  • Clinical Adoption: Pushing CI techniques from successful medical image analysis into actual clinical practice.
  • Generalization: Overcoming the failure of current reinforcement learning to generalize across different tasks.
  • Stability: Ensuring that self-learning, adaptive systems remain mathematically stable and predictable in real-world engineering applications.

Conclusion

The 2020 CIS Awards underscore a clear trajectory for the field: the integration of evolutionary principles, the development of internalized goal-setting, and the use of complex simulations to bridge the gap between theory and practice. These contributions provide the roadmap for the next generation of robust, general-purpose artificial intelligence.

Summary updated

BibTeX
@article{alippi2020ieee,
  author = {Cesare Alippi},
  title = {{2020 IEEE CIS Awards [Society Briefs]}},
  journal = {IEEE Comput. Intell. Mag.},
  year = {2020},
  doi = {10.1109/mci.2019.2954636},
}
Figure