Cost-aware learning of relevant contextual variables within Bayesian optimization

Contextual Bayesian Optimization (CBO) is a powerful framework for optimizing black-box, expensive-to-evaluate functions with respect to design variables, while simultaneously efficiently integrating relevant contextual information regarding the environment, such as experimental conditions. However, in many practical scenarios, the relevance of contextual variables is not necessarily known beforehand. Moreover, the contextual variables can sometimes be optimized themselves, a setting that current CBO algorithms do not take into account. Optimizing contextual variables may be costly, which raises the question of determining a minimal relevant subset. In this paper, we frame this problem as a cost-aware model selection BO task and address it using a novel method, Sensitivity-Analysis-Driven Contextual BO (SADCBO). We learn the relevance of context variables by sensitivity analysis of the posterior surrogate model at specific input points, whilst minimizing the cost of optimization by leveraging recent developments on early stopping for BO. We empirically evaluate our proposed SADCBO against alternatives on synthetic experiments together with extensive ablation studies, and demonstrate a consistent improvement across examples.

Multi-Fidelity Bayesian Optimization with Unreliable Information Sources

Bayesian optimization (BO) is a powerful framework for optimizing black-box, expensive-to-evaluate functions. Over the past decade, many algorithms have been proposed to integrate cheaper, lower-fidelity approximations of the objective function into the optimization process, with the goal of converging towards the global optimum at a reduced cost. This task is generally referred to as multi-fidelity Bayesian optimization (MFBO). However, MFBO algorithms can lead to higher optimization costs than their vanilla BO counterparts, especially when the low-fidelity sources are poor approximations of the objective function, therefore defeating their purpose. To address this issue, we propose rMFBO (robust MFBO), a methodology to make any GP-based MFBO scheme robust to the addition of unreliable information sources. rMFBO comes with a theoretical guarantee that its performance can be bound to its vanilla BO analog, with high controllable probability. We demonstrate the effectiveness of the proposed methodology on a number of numerical benchmarks, outperforming earlier MFBO methods on unreliable sources. We expect rMFBO to be particularly useful to reliably include human experts with varying knowledge within BO processes.

Reactmine: a search algorithm for inferring chemical reaction networks from time series data

Inferring chemical reaction networks (CRN) from time series data is a challenge encouraged by the growing availability of quantitative temporal data at the cellular level. This motivates the design of algorithms to infer the preponderant reactions between the molecular species observed in a given biochemical process, and help to build CRN model structure and kinetics. Existing ODE-based inference methods such as SINDy resort to least square regression combined with sparsity-enforcing penalization, such as Lasso. However, when the input time series are only available in wild type conditions in which all reactions are present, we observe that current methods fail to learn sparse models. Results: We present Reactmine, a CRN learning algorithm which enforces sparsity by inferring reactions in a sequential fashion within a search tree of bounded depth, ranking the inferred reaction candidates according to the variance of their kinetics, and re-optimizing the CRN kinetic parameters on the whole trace in a final pass to rank the inferred CRN candidates. We first evaluate its performance on simulation data from a benchmark of hidden CRNs, together with algorithmic hyperparameter sensitivity analyses, and then on two sets of real experimental data: one from protein fluorescence videomicroscopy of cell cycle and circadian clock markers, and one from biomedical measurements of systemic circadian biomarkers possibly acting on clock gene expression in peripheral organs. We show that Reactmine succeeds both on simulation data by retrieving hidden CRNs where SINDy fails, and on the two real datasets by inferring reactions in agreement with previous studies.