580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How do different encoding schemes and genetic operators affect the performance of real-coded genetic algorithms in continuous optimization problems?
This research area investigates the design and selection of encoding schemes—particularly real-coded versus binary-coded—and associated genetic operators such as crossover and mutation, focusing on their impact on exploration, exploitation, precision, convergence speed, and robustness when solving continuous parameter optimization problems. Understanding these factors is crucial for improving the efficacy of genetic algorithms in high-dimensional, continuous domains common in engineering and AI.
Key finding: The paper provides a comprehensive comparative review of real-coded genetic algorithms (RCGAs), highlighting that real-number representation—where chromosomes are vectors of floating-point numbers—is particularly natural and... Read more
Key finding: This study proposes a novel hybrid GA that divides computational effort between a binary-coded GA (bGA) controlling coarse exploration to identify promising regions and a real-coded GA (rGA) conducting fine-grained... Read more
Key finding: The paper develops a modified real-coded GA featuring biased initialization, dynamic parameter tuning, and elitism to address parameter identification tasks involving nonlinear systems modeled by differential equations. The... Read more
A System on Chip Development of Customizable GA Architecture for Real Parameter Optimization Problem
Key finding: This work presents a hardware-implemented real-coded GA architecture optimized for continuous parameter optimization problems, emphasizing a modular design balancing real-time performance with flexible reconfigurability.... Read more
2. How can genetic algorithms be effectively applied to cryptanalysis and coding theory problems involving code decoding and code construction?
This theme explores the novel application of genetic algorithms (GAs) for solving complex problems in cryptanalysis—such as decoding ciphers like Vigenère—and in coding theory, including the construction and decoding of error-correcting codes over various algebraic structures. The focus is on designing GA schemes that leverage specific problem structures (e.g., codeword properties, fitness functions based on language statistics or codeword weights) to efficiently approximate solutions that are otherwise computationally intensive or NP-hard, thereby demonstrating GAs as valuable tools in information theory and security.
Key finding: The study employs a GA to successfully break Vigenère ciphertexts by representing keyword guesses as chromosomes and using a fitness function based on counts of valid English words in decoded text. Introducing an... Read more
Key finding: This paper presents a GA-based soft-decision decoding algorithm for linear block codes, formulating decoding as an optimization problem minimizing Euclidean distance between received real-valued vectors and candidate... Read more
Key finding: The authors develop a computational methodology leveraging GAs and algebraic insights to construct quasi-twisted (QT) two-weight codes, which have exactly two distinct nonzero codeword weights. Notably, the paper generalizes... Read more
Key finding: This work presents a novel database compiling best known linear codes over the ring Z_4, which are of significant interest due to their richer algebraic structure behind quaternary coding. The authors develop genetic code... Read more
3. What are the advances in algorithmic and hardware frameworks to improve the efficiency and reliability of genetic algorithms and coded computation in practical engineering and information systems?
This theme covers the development of algorithmic frameworks and hardware implementations designed to optimize the computational efficiency, real-time performance, and fault tolerance of genetic algorithms and coded computation approaches. It includes the creation of modular architectures for real-coded GAs on system-on-chip platforms, as well as novel coded computation paradigms that leverage error-correcting codes with decoding and re-encoding steps to enhance reliability and performance in noisy environments, particularly for applications in areas like electromagnetic interference shielding design and neural networks.
Key finding: This paper introduces computation coding, a technique decomposing any given matrix into a product of codebook and wiring matrices consisting of entries that are zero or signed powers of two. The method drastically reduces the... Read more
A System on Chip Development of Customizable GA Architecture for Real Parameter Optimization Problem
Key finding: This work showcases a modular GA architecture implemented on an FPGA that supports real-parameter optimization with minimal hardware utilization. The system achieves a balance between real-time performance and flexibility,... Read more
Key finding: The study applies a real-coded GA with simulated binary crossover and parameter-based mutation to design multi-layer electromagnetic shields achieving target shielding effectiveness under X-band frequencies. By formulating... Read more
Key finding: This paper extends and refines Spielman’s framework for performing computation directly on encoded data protected by error-correcting codes. The generalized model introduces flexibility in code choices and encoding alphabets,... Read more