CyberSec Research

Developmental changes in body morphology profoundly shape locomotion in animals, yet artificial agents and robots are typically trained under static physical parameters. Inspired by ontogenetic scaling of muscle power in biology, we propose Scaling Mechanical Output over Lifetime (SMOL), a novel curriculum that dynamically modulates robot actuator strength to mimic natural variations in power-to-weight ratio during growth and ageing. Integrating SMOL into the MAP-Elites quality-diversity framework, we vary the torque in standard robotics tasks to mimic the evolution of strength in animals as they grow up and as their body changes. Through comprehensive empirical evaluation, we show that the SMOL schedule consistently elevates both performance and diversity of locomotion behaviours across varied control scenarios, by allowing agents to leverage advantageous physics early on to discover skills that act as stepping stones when they reach their final standard body properties. Based on studies of the total power output in humans, we also implement the SMOL-Human schedule that models isometric body variations due to non-linear changes like puberty, and study its impact on robotics locomotion.

This paper presents a machine learning-based approach to correct inference errors caused by stuck-at faults in fully analog ReRAM-based neuromorphic circuits. Using a Design-Technology Co-Optimization (DTCO) simulation framework, we model and analyze six spatial defect types-circular, circular-complement, ring, row, column, and checkerboard-across multiple layers of a multi-array neuromorphic architecture. We demonstrate that the proposed correction method, which employs a lightweight neural network trained on the circuit's output voltages, can recover up to 35% (from 55% to 90%) inference accuracy loss in defective scenarios. Our results, based on handwritten digit recognition tasks, show that even small corrective networks can significantly improve circuit robustness. This method offers a scalable and energy-efficient path toward enhanced yield and reliability for neuromorphic systems in edge and internet-of-things (IoTs) applications. In addition to correcting the specific defect types used during training, our method also demonstrates the ability to generalize-achieving reasonable accuracy when tested on different types of defects not seen during training. The framework can be readily extended to support real-time adaptive learning, enabling on-chip correction for dynamic or aging-induced fault profiles.

Despite significant efforts to manually design high-performance evolutionary algorithms, their adaptability remains limited due to the dynamic and ever-evolving nature of real-world problems. The "no free lunch" theorem highlights that no single algorithm performs optimally across all problems. While online adaptation methods have been proposed, they often suffer from inefficiency, weak convergence, and limited generalization on constrained optimization problems (COPs). To address these challenges, we introduce a novel framework for automated component configuration in Differential Evolution (DE) algorithm to address COPs, powered by Deep Reinforcement Learning (DRL). Specifically, we propose SuperDE, a foundation model that dynamically configures DE's evolutionary components based on real-time evolution. Trained offline through meta-learning across a wide variety of COPs, SuperDE is capable of recommending optimal per-generation configurations for unseen problems in a zero-shot manner. Utilizing a Double Deep Q-Network (DDQN), SuperDE adapts its configuration strategies in response to the evolving population states during optimization. Experimental results demonstrate that SuperDE significantly outperforms existing state-of-the-art algorithms on benchmark test suites, achieving superior generalization and optimization performance.

Efficient planning and sequence selection are central to intelligence, yet current approaches remain largely incompatible with biological computation. Classical graph algorithms like Dijkstra's or A* require global state and biologically implausible operations such as backtracing, while reinforcement learning methods rely on slow gradient-based policy updates that appear inconsistent with rapid behavioral adaptation observed in natural systems. We propose a biologically plausible algorithm for shortest-path computation that operates through local spike-based message-passing with realistic processing delays. The algorithm exploits spike-timing coincidences to identify nodes on optimal paths: Neurons that receive inhibitory-excitatory message pairs earlier than predicted reduce their response delays, creating a temporal compression that propagates backwards from target to source. Through analytical proof and simulations on random spatial networks, we demonstrate that the algorithm converges and discovers all shortest paths using purely timing-based mechanisms. By showing how short-term timing dynamics alone can compute shortest paths, this work provides new insights into how biological networks might solve complex computational problems through purely local computation and relative spike-time prediction. These findings open new directions for understanding distributed computation in biological and artificial systems, with possible implications for computational neuroscience, AI, reinforcement learning, and neuromorphic systems.

The educational competition optimizer is a recently introduced metaheuristic algorithm inspired by human behavior, originating from the dynamics of educational competition within society. Nonetheless, ECO faces constraints due to an imbalance between exploitation and exploration, rendering it susceptible to local optima and demonstrating restricted effectiveness in addressing complex optimization problems. To address these limitations, this study presents an enhanced educational competition optimizer (IECO-MCO) utilizing multi-covariance learning operators. In IECO, three distinct covariance learning operators are introduced to improve the performance of ECO. Each operator effectively balances exploitation and exploration while preventing premature convergence of the population. The effectiveness of IECO is assessed through benchmark functions derived from the CEC 2017 and CEC 2022 test suites, and its performance is compared with various basic and improved algorithms across different categories. The results demonstrate that IECO-MCO surpasses the basic ECO and other competing algorithms in convergence speed, stability, and the capability to avoid local optima. Furthermore, statistical analyses, including the Friedman test, Kruskal-Wallis test, and Wilcoxon rank-sum test, are conducted to validate the superiority of IECO-MCO over the compared algorithms. Compared with the basic algorithm (improved algorithm), IECO-MCO achieved an average ranking of 2.213 (2.488) on the CE2017 and CEC2022 test suites. Additionally, the practical applicability of the proposed IECO-MCO algorithm is verified by solving constrained optimization problems. The experimental outcomes demonstrate the superior performance of IECO-MCO in tackling intricate optimization problems, underscoring its robustness and practical effectiveness in real-world scenarios.

Metaheuristics are widely applied for their ability to provide more efficient solutions. The RIME algorithm is a recently proposed physical-based metaheuristic algorithm with certain advantages. However, it suffers from rapid loss of population diversity during optimization and is prone to fall into local optima, leading to unbalanced exploitation and exploration. To address the shortcomings of RIME, this paper proposes a modified RIME with covariance learning and diversity enhancement (MRIME-CD). The algorithm applies three strategies to improve the optimization capability. First, a covariance learning strategy is introduced in the soft-rime search stage to increase the population diversity and balance the over-exploitation ability of RIME through the bootstrapping effect of dominant populations. Second, in order to moderate the tendency of RIME population to approach the optimal individual in the early search stage, an average bootstrapping strategy is introduced into the hard-rime puncture mechanism, which guides the population search through the weighted position of the dominant populations, thus enhancing the global search ability of RIME in the early stage. Finally, a new stagnation indicator is proposed, and a stochastic covariance learning strategy is used to update the stagnant individuals in the population when the algorithm gets stagnant, thus enhancing the ability to jump out of the local optimal solution. The proposed MRIME-CD algorithm is subjected to a series of validations on the CEC2017 test set, the CEC2022 test set, and the experimental results are analyzed using the Friedman test, the Wilcoxon rank sum test, and the Kruskal Wallis test. The results show that MRIME-CD can effectively improve the performance of basic RIME and has obvious superiorities in terms of solution accuracy, convergence speed and stability.

Numerous purportedly improved metaheuristics claim superior performance based on equivalent function evaluations (FEs), yet often conceal additional computational burdens in more intensive iterations, preprocessing stages, or hyperparameter tuning. This paper posits that wall-clock time, rather than solely FEs, should serve as the principal budgetary constraint for equitable comparisons. We formalize a fixed-time, restart-fair benchmarking protocol wherein each algorithm is allotted an identical wall-clock time budget per problem instance, permitting unrestricted utilization of restarts, early termination criteria, and internal adaptive mechanisms. We advocate for the adoption of anytime performance curves, expected running time (ERT) metrics, and performance profiles that employ time as the cost measure, all aimed at predefined targets. Furthermore, we introduce a concise, reproducible checklist to standardize reporting practices and mitigate undisclosed computational overheads. This approach fosters more credible and practically relevant evaluations of metaheuristic algorithms.

A machine learning method is proposed using two agents that simulate the biological behavior of a predator and a prey. In this method, the predator and the prey interact with each other - the predator chases the prey while the prey runs away from the predator - to perform an optimization on the landscape. This method allows, for the case of a ravine landscape (i.e., a landscape with narrow ravines and with gentle slopes along the ravines) to avoid getting optimization stuck in the ravine. For this, in the optimization over a ravine landscape the predator drives the prey along the ravine. Thus we also call this approach, for the case of ravine landscapes, the driven hunt method. For some examples of grokking (i.e., delayed generalization) problems we show that this method allows for achieving up to a hundred times faster learning compared to the standard learning procedure.

Existing key-value (KV) cache compression methods typically rely on heuristics, such as uniform cache allocation across layers or static eviction policies, however, they ignore the critical interplays among layer-specific feature patterns and task performance, which can lead to degraded generalization. In this paper, we propose EvolKV, an adaptive framework for layer-wise, task-driven KV cache compression that jointly optimizes the memory efficiency and task performance. By reformulating cache allocation as a multi-objective optimization problem, EvolKV leverages evolutionary search to dynamically configure layer budgets while directly maximizing downstream performance. Extensive experiments on 11 tasks demonstrate that our approach outperforms all baseline methods across a wide range of KV cache budgets on long-context tasks and surpasses heuristic baselines by up to 7 percentage points on GSM8K. Notably, EvolKV achieves superior performance over the full KV cache setting on code completion while utilizing only 1.5% of the original budget, suggesting the untapped potential in learned compression strategies for KV cache budget allocation.

Large Language Models (LLMs), with their strong understanding and reasoning capabilities, are increasingly being explored for tackling optimization problems, especially in synergy with evolutionary computation. Despite rapid progress, however, the field still lacks a unified synthesis and a systematic taxonomy. This survey addresses this gap by providing a comprehensive review of recent developments and organizing them within a structured framework. We classify existing research into two main stages: LLMs for optimization modeling and LLMs for optimization solving. The latter is further divided into three paradigms according to the role of LLMs in the optimization workflow: LLMs as stand-alone optimizers, low-level LLMs embedded within optimization algorithms, and high-level LLMs for algorithm selection and generation. For each category, we analyze representative methods, distill technical challenges, and examine their interplay with traditional approaches. We also review interdisciplinary applications spanning the natural sciences, engineering, and machine learning. By contrasting LLM-driven and conventional methods, we highlight key limitations and research gaps, and point toward future directions for developing self-evolving agentic ecosystems for optimization. An up-to-date collection of related literature is maintained at https://github.com/ishmael233/LLM4OPT.

Artifacts in electroencephalography (EEG) -- muscle, eye movement, electrode, chewing, and shiver -- confound automated analysis yet are costly to label at scale. We study whether modern generative models can synthesize realistic, label-aware artifact segments suitable for augmentation and stress-testing. Using the TUH EEG Artifact (TUAR) corpus, we curate subject-wise splits and fixed-length multi-channel windows (e.g., 250 samples) with preprocessing tailored to each model (per-window min-max for adversarial training; per-recording/channel $z$-score for diffusion). We compare a conditional WGAN-GP with a projection discriminator to a 1D denoising diffusion model with classifier-free guidance, and evaluate along three axes: (i) fidelity via Welch band-power deltas ($\Delta\delta,\ \Delta\theta,\ \Delta\alpha,\ \Delta\beta$), channel-covariance Frobenius distance, autocorrelation $L_2$, and distributional metrics (MMD/PRD); (ii) specificity via class-conditional recovery with lightweight $k$NN/classifiers; and (iii) utility via augmentation effects on artifact recognition. In our setting, WGAN-GP achieves closer spectral alignment and lower MMD to real data, while both models exhibit weak class-conditional recovery, limiting immediate augmentation gains and revealing opportunities for stronger conditioning and coverage. We release a reproducible pipeline -- data manifests, training configurations, and evaluation scripts -- to establish a baseline for EEG artifact synthesis and to surface actionable failure modes for future work.

Spiking neural networks offer a promising path toward energy-efficient, brain-like associative memory. This paper introduces Word2Spike, a novel rate coding mechanism that combines continuous word embeddings and neuromorphic architectures. We develop a one-to-one mapping that converts multi-dimensional word vectors into spike-based attractor states using Poisson processes. Using BitNet b1.58 quantization, we maintain 97% semantic similarity of continuous embeddings on SimLex-999 while achieving 100% reconstruction accuracy on 10,000 words from OpenAI's text-embedding-3-large. We preserve analogy performance (100% of original embedding performance) even under intentionally introduced noise, indicating a resilient mechanism for semantic encoding in neuromorphic systems. Next steps include integrating the mapping with spiking transformers and liquid state machines (resembling Hopfield Networks) for further evaluation.

Gradient-based neural network training traditionally enforces symmetry between forward and backward propagation, requiring activation functions to be differentiable (or sub-differentiable) and strictly monotonic in certain regions to prevent flat gradient areas. This symmetry, linking forward activations closely to backward gradients, significantly restricts the selection of activation functions, particularly excluding those with substantial flat or non-differentiable regions. In this paper, we challenge this assumption through mathematical analysis, demonstrating that precise gradient magnitudes derived from activation functions are largely redundant, provided the gradient direction is preserved. Empirical experiments conducted on foundational architectures - such as Multi-Layer Perceptrons (MLPs), Convolutional Neural Networks (CNNs), and Binary Neural Networks (BNNs) - confirm that relaxing forward-backward symmetry and substituting traditional gradients with simpler or stochastic alternatives does not impair learning and may even enhance training stability and efficiency. We explicitly demonstrate that neural networks with flat or non-differentiable activation functions, such as the Heaviside step function, can be effectively trained, thereby expanding design flexibility and computational efficiency. Further empirical validation with more complex architectures remains a valuable direction for future research.

Aiming at the shortcomings of the gazelle optimization algorithm, such as the imbalance between exploration and exploitation and the insufficient information exchange within the population, this paper proposes a multi-strategy improved gazelle optimization algorithm (MSIGOA). To address these issues, MSIGOA proposes an iteration-based updating framework that switches between exploitation and exploration according to the optimization process, which effectively enhances the balance between local exploitation and global exploration in the optimization process and improves the convergence speed. Two adaptive parameter tuning strategies improve the applicability of the algorithm and promote a smoother optimization process. The dominant population-based restart strategy enhances the algorithms ability to escape from local optima and avoid its premature convergence. These enhancements significantly improve the exploration and exploitation capabilities of MSIGOA, bringing superior convergence and efficiency in dealing with complex problems. In this paper, the parameter sensitivity, strategy effectiveness, convergence and stability of the proposed method are evaluated on two benchmark test sets including CEC2017 and CEC2022. Test results and statistical tests show that MSIGOA outperforms basic GOA and other advanced algorithms. On the CEC2017 and CEC2022 test sets, the proportion of functions where MSIGOA is not worse than GOA is 92.2% and 83.3%, respectively, and the proportion of functions where MSIGOA is not worse than other algorithms is 88.57% and 87.5%, respectively. Finally, the extensibility of MSIGAO is further verified by several engineering design optimization problems.

Spiking Neural Networks (SNNs) are promising for neuromorphic computing due to their biological plausibility and energy efficiency. However, training methods like Backpropagation Through Time (BPTT) and Real Time Recurrent Learning (RTRL) remain computationally intensive. This work introduces an integer-only, online training algorithm using a mixed-precision approach to improve efficiency and reduce memory usage by over 60%. The method replaces floating-point operations with integer arithmetic to enable hardware-friendly implementation. It generalizes to Convolutional and Recurrent SNNs (CSNNs, RSNNs), showing versatility across architectures. Evaluations on MNIST and the Spiking Heidelberg Digits (SHD) dataset demonstrate that mixed-precision models achieve accuracy comparable to or better than full-precision baselines using 16-bit shadow and 8- or 12-bit inference weights. Despite some limitations in low-precision and deeper models, performance remains robust. In conclusion, the proposed integer-only online learning algorithm presents an effective solution for efficiently training SNNs, enabling deployment on resource-constrained neuromorphic hardware without sacrificing accuracy.

Mathematical morphology provides a nonlinear framework for image and spatial data processing and analysis. Although there have been many successful applications of mathematical morphology to vector-valued images, such as color and hyperspectral images, there is still no consensus on the most suitable vector ordering for constructing morphological operators. This paper addresses this issue by examining a reduced ordering approximating the Condorcet ranking derived from a set of vector orderings. Inspired by voting problems, the Condorcet ordering ranks elements from most to least voted, with voters representing different orderings. In this paper, we develop a machine learning approach that learns a reduced ordering that approximates the Condorcet ordering. Preliminary computational experiments confirm the effectiveness of learning the reduced mapping to define vector-valued morphological operators for color images.

Swarm intelligence algorithms have demonstrated remarkable success in solving complex optimization problems across diverse domains. However, their widespread adoption is often hindered by limited transparency in how algorithmic components influence performance. This work presents a multi-faceted investigation of Particle Swarm Optimization (PSO) to further understand the key role of different topologies for better interpretability and explainability. To achieve this objective, we first develop a comprehensive landscape characterization framework using Exploratory Landscape Analysis (ELA) to quantify problem difficulty and identify critical features affecting the optimization performance of PSO. Next, we conduct a rigorous empirical study comparing three fundamental swarm communication architectures -- Ring, Star, and Von Neumann topologies -- analysing their distinct impacts on exploration-exploitation balance, convergence behaviour, and solution quality and eventually develop an explainable benchmarking framework for PSO, to decode how swarm topologies affects information flow, diversity, and convergence. Based on this, a novel machine learning approach for automated algorithm configuration is introduced for training predictive models on extensive Area over the Convergence Curve (AOCC) data to recommend optimal settings based on problem characteristics. Through systematic experimentation across twenty four benchmark functions in multiple dimensions, we establish practical guidelines for topology selection and parameter configuration. These findings advance the development of more transparent and reliable swarm intelligence systems. The source codes of this work can be accessed at https://github.com/GitNitin02/ioh_pso.

Continual learning, the ability to acquire and transfer knowledge through a models lifetime, is critical for artificial agents that interact in real-world environments. Biological brains inherently demonstrate these capabilities while operating within limited energy and resource budgets. Achieving continual learning capability in artificial systems considerably increases memory and computational demands, and even more so when deploying on platforms with limited resources. In this work, Genesis, a spiking continual learning accelerator, is proposed to address this gap. The architecture supports neurally inspired mechanisms, such as activity-dependent metaplasticity, to alleviate catastrophic forgetting. It integrates low-precision continual learning parametersand employs a custom data movement strategy to accommodate the sparsely distributed spikes. Furthermore, the architecture features a memory mapping technique that places metaplasticity parameters and synaptic weights in a single address location for faster memory access. Results show that the mean classification accuracy for Genesis is 74.6% on a task-agnostic split-MNIST benchmark with power consumption of 17.08mW in a 65nm technology node.