Time-series forecasting is widely applied across various domains, yet most approaches rely on predefined time steps given by each problem. Based on observations from dynamic systems with known ground truth, we identify that large-step forecasts can lead to substantial errors due to insufficient modeling of continuous dynamics. To address this, we propose a micro-step time-series regression technique that decomposes predictions into smaller intervals, so that genetic programming-based feature construction can capture finer temporal patterns to improve the prediction performance. Specifically, we employ linear interpolation to allow the evolutionary feature construction process to learn from incremental changes, reducing the difficulty of time-series regression. Experiments on 100 datasets from the M4 forecasting benchmark demonstrate that micro-step regression significantly enhances prediction accuracy compared to traditional methods using raw time steps. Further analysis reveals that features trained on micro-step data evolve into simpler structures, promoting both generalization and interpretability.

To ensure soil preservation, it is essential to incorporate the soil's ability to provide ecosystem services into the spatial planning process. For well-informed planning decisions, stakeholders need spatially explicit information on the state of the soils and the functions they fulfil, with sufficient spatial resolution and quantified uncertainty. It has been shown that Digital Soil Mapping (DSM) products can provide such information. However, in some cases, fine spatial resolution coupled with high levels of uncertainty may lead stakeholders to overlook the inherent uncertainties in the information. Spatial aggregation of DSM products opens up a promising avenue for obtaining maps that are more tailored to the users' scales of decision making while facilitating uncertainty communication. In this perspective, we propose a new spatial aggregation approach relying on spatially constrained agglomerative clustering (AC). The spatial aggregation approach is applied to a 25-m-resolution soil potential multifunctionality index (SPMI) map developed for the coastal plain of the Occitanie Region. This DSM product was increasingly aggregated to obtain SPMI maps of different resolutions displaying two distinct areal metrics: proportions of area above a given threshold of SPMI, and mean SPMI. Each map was evaluated through a set of indicators selected for their potential impact on user decision making: mean spatial resolution, overall predicted uncertainty, quantity of information and mean within-unit variability. The maps were compared with respect to these indicators to other maps obtained with alternative aggregation methods employed in DSM literature (maps aggregated according to some administrative units and QuadMaps). We show that all the tested aggregation methods produced a substantial decrease of the map uncertainty with moderate loss of spatial resolution. However, only AC preserved the fine spatial pattern of the initial DSM product while enabling fine tuning of the uncertainty displayed to end-users. We show that AC can simplify the identification of extensive regions characterized by low uncertainty without losing information regarding soil multifunctionality, thereby facilitating and enhancing the efficiency of planning decisions.

The mathematical theory of viability, developed to formalize problems related to natural and social phenomena, investigates the evolution of dynamical systems under constraints. A main objective of this theory is to design control laws to keep systems inside viable domains. Control laws are traditionally defined as rules, based on the current position in the state space with respect to the boundaries of the viability kernel. However, finding these boundaries is a computationally expensive procedure, feasible only for trivial systems. We propose an approach based on Genetic Programming (GP) to discover control laws for viability problems in analytic form. Such laws could keep a system viable without the need of computing its viability kernel, facilitate communication with stakeholders, and improve explainability. A candidate set of control rules is encoded as GP trees describing equations. Evaluation is noisy, due to stochastic sampling: initial conditions are randomly drawn from the state space of the problem, and for each, a system of differential equations describing the system is solved, creating a trajectory. Candidate control laws are rewarded for keeping viable as many trajectories as possible, for as long as possible. The proposed approach is evaluated on established benchmarks for viability and delivers promising results.

Many studies aim to create fully automatic classification pipelines where human input is minimal or non-existent. However, for certain classification problems, where data is scarce or difficult to interpret, utilising the human in the loop (HITL) can improve classification performance. Furthermore, combining traditional classification methods with a state-of-the-art convolutional neural network (CNN) may achieve good performance. Our study therefore aims to provide an ensemble method that outperforms a CNN and traditional classifiers, and to create an iterative HITL pipeline to automatically detect peas from magnetic resonance imaging (MRI) images. Our ensemble contains one CNN (AlexNet), and two traditional classifiers that take radiomic data (Quadratic Discriminant Analysis (QDA) and a support vector machine (SVM)). At each iteration, the humans in the loop correct the results of the previous classification. This data is then added to our training and testing data to improve the pea detection at the next iteration. In most cases our ensemble performs better than the CNN or traditional classifiers alone but is not a perfect trend and in some of the cases the CNN performs marginally better. Overall, we found that the addition of HITL improves performance compared to without.

In this paper, we present a test of an interactive modelling scheme in real conditions. The aim is to use this scheme to identify the physiological responses of microorganisms at different scales in a real industrial application context. The originality of the proposed tool, Biosys-LiDeOGraM, is to generate through a human–machine cooperation a consistent and concise model from molecules to microbial population scales: If multi-omics measurements can be connected relatively easily to the response of the biological system at the molecular scale, connecting them to the macroscopic level of the biosystem remains a difficult task, where human knowledge plays a crucial role. The use-case considered here pertains to an engineering process of freeze-drying and storage of Lactic Acid Bacteria. Producing a satisfying model of this process is a challenge due to (i) the scarcity and variability of the experimental dataset, (ii) the complexity and multi-scale nature of biological phenomena, and (iii) the wide knowledge about the biological mechanisms involved in this process. The Biosys-LiDeOGraM tool has two main components that can have to be utilized in an iterative manner: the Genomic Interactive Clustering (GIC) module and the Interactive Multi-Scale modellIng Exploration (IMSIE) module, both involve users in their learning loops. Applying our approach to a dataset of 2,741 genes, an initial model, as a graph involving 33 variables and 165 equations, was first built. Then the system was able to interactively improve a synthetic version of this model using only 27 variables and 16 equations. The final graph providing a consistent and explainable biological model. This graphical representation allows various user interpretations at local and global scales, an easy confrontation with data, and an exploration of various assumptions. Finally Biosys-LiDeOGraM is easily transferable to other use-cases of multi-scale modelling using ‘functional’ graphs. Author summary The use of “omics” data for understanding biological systems has become prevalent in several research domains. However, the data generated from diverse macroscopic scales used for this purpose is highly heterogeneous and challenging to integrate. Yet, it is crucial to incorporate this information to gain a comprehensive understanding of the underlying biological system. Although various integrative analysis methods that have been developed provide predictive molecular-scale models, they only offer a mechanistic view of the biological system at the cellular level. In addition, they often focus on specific biological hypotheses through dedicated case studies, making it difficult to apply their results to other scientific problems. To address these issues, we propose an interactive multi-scale modelling approach to integrate cross-scale relationships providing predictive and potentially explanatory models. A proof-of-concept tool has been developed and was validated in the context of the bioproduction of Lactococcus lactis , a bacterial species of high economic interest in the food industry and for which the control of the bioprocess is essential to guarantee its viability and functionality. Our approach can be applied to any biological system that can be defined through a set of variables, constraints and scales.

Particle Swarm Optimisation (PSO) and Evolutionary Algorithms (EAs) differ in various ways, in particular with respect to information sharing and diversity management, making their scopes of applications very diverse. Combining the advantages of both approaches is very attractive and has been successfully achieved through hybridisation. Another possible improvement, notably for addressing scalability issues, is cooperation. It has first been developed for co-evolution in EA techniques and it is now used in PSO. However, until now, attempts to make PSO cooperate have been based on multi-population schemes almost exclusively. The focus of this paper is set on single-population schemes, or fine-grained cooperation. By analogy with an evolutionary scheme that has long been proved effective, the fly algorithm (FA), we design and compare a cooperative PSO (coPSO), and a PSO-flavoured fly algorithm. Experiments run on a benchmark, the Lamp problem, show that fine-grained cooperation based on marginal fitness evaluations and steady-state schemes outperforms classical techniques when the dimension of the problem increases. These preliminary results highlight interesting future directions of research on fine-grained cooperation schemes, by combining features of PSO and FA.

Scattering studies of milk and milk products, which are highly relevant food products on the global market, are often utilized and reported in literature to investigate and understand the subtle microscopic structural differences between dairy samples. These structural features determine the physical properties and ultimately the texture of milk products and, thus, also influence the consumer's experience. Small-angle neutron scattering is a prominent example, which enables observations of length scales, which convey proteins and fat globules in food-grade milk. In addition, deuteration enables contrast variations between the constituents of dairy products. In this study, we investigate the potential of probing small-angle neutron scattering from milk samples through quantitative neutron dark-field imaging using grating interferometry, to establish the feasibility of studying, in particular, fat globules and milk gel structures with this spatially resolved scattering technique.

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

CONTEXT: So far, digital technology development in agriculture has mainly dealt with precision agriculture, often associated with conventional large-scale systems. The emergence of digital agriculture -based on the triptych of "new data sources / new processing methods / new inter-connection capacities (internet)" -opens up prospects for mobilizing digital technologies to accelerate the deployment of other forms of agriculture, such as agroecology. A specific research agenda must therefore be built to redirect researchers specialized in digital technologies towards these new issues. This construction is significant because digital technology and agro-ecology are disruptive innovations that shake up the actors' practices, agricultural innovation ecosystems, and value chains.OBJECTIVE: An interdisciplinary group of INRAE researchers (covering 10 scientific departments) was mandated to carry out this reflection, with the objective of developing a research agenda to better couple digitalization and agroecology, in order to pave the way for responsible digital farming. The group used the framework of responsible research and innovation.METHOD: Over 18 months, the group met monthly by video-conference, to overcome the interdisciplinarity barrier, and at three face-to-face seminars, where creative design exercises were carried out (based on a world cafe acute accent format, and "remember the future" method). This work gave rise to three prospective lines of research aimed at putting digitalization at the service of agroecology and local food systems. These topics prioritize research that fosters innovations in digital technology, as well as organisations and policies that (1) accelerate the agroeco-logical transition on the farm and in the territories, (2) manage the territories as commons, (3) empower farmers and consumers. Then, the group examined these three prospective lines of research from an RRI perspective as well as three current research topics on digital agriculture (digital soil mapping, precision agriculture, tech-nologies for food wastage reduction).RESULTS AND CONCLUSIONS: This work allowed us to highlight the gaps between current research on digital agriculture and the RRI expectations, and the tensions (between rationalization and diversity of farming systems, between complexity of agroecological systems and the need for simplification of models, and finally between data speculation and frugality). We were also able to refine the specific scientific questions of each prospective line of research and finally to draw attention to the key levers that will have to be integrated if these research efforts are to be approached from an RRI perspective.SIGNIFICANCE: This contribution shows RRI can be used not only to reflect on research practices but also as a framework to build a research agenda paving the way for responsible digital agriculture.

Numerous studies, including our previous work with lemon juice, have reported that low-pH meals reduce the glycemic response to starchy foods. However, the underlying mechanism is not yet understood. Tea, for its polyphenol content, has also been investigated. The main objective of this research was to concurrently study gastric emptying, appetite perceptions and glycemic responses to bread consumed with water, tea, or lemon juice. In this randomized, crossover intervention, ten participants consumed equal portions of bread (100 g) with 250 mL of water, water-diluted lemon juice, or black tea at breakfast. Gastric volumes, blood glucose concentrations and appetite perceptions were alternately assessed over 180 min using magnetic resonance imaging, the finger-prick method and visual analogue scales, respectively. Compared to water, lemon juice led to a 1.5 fold increase of the volume of gastric contents, 30 min after the meal (454.0 ± 18.6 vs. 298.4 ± 19.5 mL, $$\overline{x}$$  ± SEM P < 0.00001). Gastric emptying was also 1.5 times faster (P < 0.01). Conversely, lemon juice elicited a lower glycemic response than water (blood glucose concentrations at t = 55 min were 35% lower, P = 0.039). Tea had no effect. Changes in appetite perceptions and gastric volumes correlated well, but with no significant differences between the meals. Lemon juice lowered the glycemic response and increased both gastric secretions and emptying rate. The results are compatible with the hypothesis that the reduction of the glycemic response is mainly due to the interruption of starch hydrolysis via the acid-inhibition of salivary α-amylase. NCT03265392, August 29, 2017.

In this paper, we demonstrate how analytic provenance can be exploited to re-construct user trust in a guided Visual Analytics (VA) system, and suggest that interaction log data analysis can be a valuable tool for on-line trust monitoring. Our approach explores objective trust measures that can be continuously tracked and updated during the exploration, and reflect both the confidence of the user in system suggestions, and the uncertainty of the system with regards to user goals. We argue that this approach is more suitable for guided VA systems such as ours, where user strategies, goals and even trust can evolve over time, in reaction to new system feedback and insights from the exploration. Through the analysis of log data from a past user study with twelve participants performing a guided visual analysis task, we found that the stability of user exploration strategies is a promising factor to study trust. However, indirect metrics based on provenance, such as user evaluation counts and disagreement rates, are alone not sufficient to study trust reliably in guided VA. We conclude with open challenges and opportunities for exploiting analytic provenance to support trust monitoring in guided VA systems.

The increasing use of search and optimisation algorithms in real-world applications presents new challenges to researchers to develop algorithms that are computationally efficient and are able to produce meaningful solutions. In this talk, I will describe two approaches that are aiming to address these challenges: interactive evolutionary metaheuristics and sequence-based hyperheuristics. These methods are designed to make use of human intelligence and machine learning to improve search and optimisation performance and to generate feasible solutions for real-world problems in the water industry and operations research problems. Specifically, I will demonstrate an interactive evolutionary algorithm (EA) system that is able to learn human preferences and embed them into the operation of an EA to improve objective and subjective performance criteria. I will then describe recent work in the use of machine learning to understand and create sequences of search operations within a hyperheuristic framework to better understand the problem-algorithm interface and improve search performance.

Feature selection is the process of choosing, or removing, features to obtain the most informative feature subset of minimal size. Such subsets are used to improve performance of machine learning algorithms and enable human understanding of the results. Approaches to feature selection in literature exploit several optimization algorithms. Multi-objective methods also have been proposed, minimizing at the same time the number of features and the error. While most approaches assess error resorting to the average of a stochastic K-fold cross-validation, comparing averages might be misleading. In this paper, we show how feature subsets with different average error might in fact be non-separable when compared using a statistical test. Following this idea, clusters of non-separable optimal feature subsets are identified. The performance in feature selection can thus be evaluated by verifying how many of these optimal feature subsets an algorithm is able to identify. We thus propose a multi-objective optimization approach to feature selection, EvoFS, with the objectives to i. minimize feature subset size, ii. minimize test error on a 10-fold cross-validation using a specific classifier, iii. maximize the analysis of variance value of the lowest-performing feature in the set. Experiments on classification datasets whose feature subsets can be exhaustively evaluated show that our approach is able to always find the best feature subsets. Further experiments on a high-dimensional classification dataset, that cannot be exhaustively analyzed, show that our approach is able to find more optimal feature subsets than state-of-the-art feature selection algorithms.

Various studies have shown that food structure has an impact on digestion kinetics. We focus here on the effects of gastric and intestinal enzymes (in-vitro digestion) on two canola seed storage proteins, napin and cruciferin. To monitor structure effect we conducted experiments on gels of these proteins at different pHs, yielding different structures and elastic modulus. What is new is to get information on the mechanisms at the lowest scales, using imaging and radiation scattering at large facilities: Synchrotron fluorescence microscopy, X-Ray scattering, at SOLEIL synchrotron, and Small-Angle Neutron Scattering, at Laboratoire Leon Brillouin reactor. We can identify the mechanisms at each step and in two distinct scale ranges, observed simultaneously, the one of the individual protein scale and the one of the structure connectivity: during gelation individual canola proteins are not deeply modified in comparison with their state in solution; larger scale gel heterogeneity appears due to connectivity or aggregation in the gastric step (up to 40 min): at short scale (large q) we see that the proteins disintegration is much slowed down in gels than in solutions, particularly in the gastric phase; at larger scales (low q), we see that the gel structure is also self-resistant to the action of the enzyme (pepsin). in the intestinal step, such kinetics differences hold until major disintegration after no more than 15 min.