Giulia Pedrielli

Accurate RNA secondary-structure prediction remains difficult despite decades of thermodynamics-based algorithms and the advent of deep-learning architectures (convolutional networks, Transformers, diffusion models). In fact, the datasets that pair RNA sequences with secondary-structure labels are often low-quality, noisy, and family-imbalanced, which limits out-of-distribution generalization and exacerbates catastrophic forgetting when new data regimes are introduced. We propose a continual-learning approach based on Lifelong Bayesian Optimization (LBO), RNAFOLBO, that treats each class of RNAs obtained from latent-space clustering as a sequential task and jointly orchestrates training and hyperparameter selection of heterogeneous models (UFold, RNA-FM, RNADiffFold), while preserving prior knowledge. Concretely, we apply LBO to 15 clusters obtained by clustering RNAStrAlign in the latent space of RNAGenesis, a model specialized in contextual representation learning and latent-space structuring, achieving a mean F 1 per cluster of 0.931 (with a range of 0.177). These results surpass the strongest one-shot baseline and mitigate forgetting without full retraining. The gains persist as additional clusters are introduced. Overall, RNAFOLBO delivers higher and more stable performance and practical scalability for integrating new RNA clusters or families, enabling more robust and transferable RNA secondary-structure prediction.

Inferring causal relations in timeseries data with delayed effects is a fundamental challenge, especially when the underlying system exhibits complex dynamics that cannot be captured by simple functional mappings. Traditional approaches often fail to produce generalized and interpretable explanations, as multiple distinct input trajectories may yield nearly indistinguishable outputs. In this work, we present ruleXplain, a framework that leverages Large Language Models (LLMs) to extract formal explanations for input-output relations in simulation-driven dynamical systems. Our method introduces a constrained symbolic rule language with temporal operators and delay semantics, enabling LLMs to generate verifiable causal rules through structured prompting. ruleXplain relies on the availability of a principled model (e.g., a simulator) that maps multivariate input time series to output time series. Within ruleXplain, the simulator is used to generate diverse counterfactual input trajectories that yield similar target output, serving as candidate explanations. Such counterfactual inputs are clustered and provided as context to the LLM, which is tasked with the generation of symbolic rules encoding the joint temporal trends responsible for the patterns observable in the output times series. A closed-loop refinement process ensures rule consistency and semantic validity. We validate the framework using the PySIRTEM epidemic simulator, mapping testing rate inputs to daily infection counts; and the EnergyPlus building energy simulator, observing temperature and solar irradiance inputs to electricity needs. For validation, we perform three classes of experiments: (1) the efficacy of the ruleset through input reconstruction; (2) ablation studies evaluating the causal encoding of the ruleset; and (3) generalization tests of the extracted rules across unseen output trends with varying phase dynamics.

In this paper we propose for the first time the hyperparameter optimization (HPO) algorithm POCAII. POCAII differs from the Hyperband and Successive Halving literature by explicitly separating the search and evaluation phases and utilizing principled approaches to exploration and exploitation principles during both phases. Such distinction results in a highly flexible scheme for managing a hyperparameter optimization budget by focusing on search (i.e., generating competing configurations) towards the start of the HPO process while increasing the evaluation effort as the HPO comes to an end. POCAII was compared to state of the art approaches SMAC, BOHB and DEHB. Our algorithm shows superior performance in low-budget hyperparameter optimization regimes. Since many practitioners do not have exhaustive resources to assign to HPO, it has wide applications to real-world problems. Moreover, the empirical evidence showed how POCAII demonstrates higher robustness and lower variance in the results. This is again very important when considering realistic scenarios with extremely expensive models to train.

Conventional population-based ODE models struggle against increased level of resolution since incorporating many states exponentially increases computational costs, and demands robust calibration for numerous hyperparameters. PySIRTEM is a spatiotemporal SEIR-based epidemic simulation platform that provides high resolution analysis of viral disease progression and mitigation. Based on the authors-developed Matlab© simulator SIRTEM , PySIRTEM ‘s modular design reflects key health processes, including infection, testing, immunity, and hospitalization, enabling flexible manipulation of transition rates. Unlike SIRTEM , PySIRTEM uses a Sequential Monte Carlo (SMC) particle filter to dynamically learn epidemiological parameters using historical COVID-19 data from several U.S. states. The improved accuracy (by orders of magnitude) make PySIRTEM ideal for informed decision-making by detecting outbreaks and fluctuations. We further demonstrate PySIRTEM ‘s usability performing a factorial analysis to assess the impact of different hyperparameter configurations on the predicted epidemic dynamics. Finally, we analyze containment scenarios with varying trends, showcasing PySIRTEM ‘s adaptability and effectiveness.

Global optimization of black-box functions is challenging in high dimensions. We introduce a conceptual adaptive random search framework, Branching Adaptive Surrogate Search Optimization (BASSO), that combines partitioning and surrogate modeling for subregion sampling. We present a finite-time analysis of BASSO, and establish conditions under which it is theoretically possible to scale to high dimensions. While we do not expect that any implementation will achieve the theoretical ideal, we experiment with several BASSO variations and discuss implications on narrowing the gap between theory and implementation. Numerical results on test problems are presented.