Search

The self-incompatibility locus (S-locus) of flowering plants displays a striking allelic diversity. How such a diversity has emerged remains unclear. In this paper, we performed numerical simulations in a finite island population genetics model to investigate how population subdivision affects the diversification process at a S-locus, given that the two-genes architecture typical of S-loci involves the crossing of a fitness valley. We show that population structure slightly reduces the parameter range allowing for the diversification of self-incompatibility haplotypes (S-haplotypes), but at the same time also increases the number of these haplotypes maintained in the whole metapopulation. This increase is partly due to a higher rate of diversification and replacement of S-haplotypes within and among demes. We also show that the two-genes architecture leads to a higher diversity in structured populations compared with a simpler genetic architecture where new S-haplotypes appear in a single mutation step. Overall, our results suggest that population subdivision can act in two opposite directions: it renders S-haplotypes diversification easier, although it also increases the risk that the self-incompatibility system is lost.

On results undertaken studies systematization is carried out and in theory - methodological positions are deep on this basis practical recommendations are offered in relation to basic directions of ecological optimization of the use of earth-resource potential erosive dangerous agrolandscape by working and development of theoretical and applied bases of the adaptive-landscape systems of agriculture, newest systems of land-tenure, aimed at the rational use and protection of agricultural land, differentiated adaptation of technological means to soil-landscape-climatic factors and economic conditions of modern agricultural.The research was aimed at identifying the latest approaches to the organization of modern land use, which will necessitate the development of farming and land use systems on an ecological and landscape basis, given the different levels of intensification and resource provision, which in turn requires the improvement of existing and development of new components and the whole complex of agricultural systems, in particular in the context of climate change, risks of water and wind erosion and desertification.Many years of domestic experience in developing the scientific basis for the organization and management of agricultural land use, taking into account the basic principles of the concept of sustainable development shows that land management design based on landscape-ecological approach to the organization of agricultural land use is quite capable of becoming a link. unified system of theory and practice of organization of ecologically balanced land use at all levels.The principles of designing the organization of the territory of erosively dangerous sloping agrolandscapes for the development of soil protection adaptive-landscape system of agriculture on sloping lands (theoretical and methodological principles of landscape-adaptive land management, methodological approaches to substantiate land management projects on a landscape-ecological basis).

Alcohol-forming fatty acyl reductases (FARs) catalyze the reduction of thioesters to alcohols and are key enzymes for microbial production of fatty alcohols. Many metabolic engineering strategies utilize FARs to produce fatty alcohols from intracellular acyl-CoA and acyl-ACP pools; however, enzyme activity, especially on acyl-ACPs, remains a significant bottleneck to high-flux production. Here, we engineer FARs with enhanced activity on acyl-ACP substrates by implementing a machine learning (ML)-driven approach to iteratively search the protein fitness landscape. Over the course of ten design-test-learn rounds, we engineer enzymes that produce over twofold more fatty alcohols than the starting natural sequences. We characterize the top sequence and show that it has an enhanced catalytic rate on palmitoyl-ACP. Finally, we analyze the sequence-function data to identify features, like the net charge near the substrate-binding site, that correlate with in vivo activity. This work demonstrates the power of ML to navigate the fitness landscape of traditionally difficult-to-engineer proteins., Fatty acyl reductases (FARs) are critical enzymes in the biosynthesis of fatty alcohols and have the ability to directly acces acyl-ACP substrates. Here, authors couple machine learning-based protein engineering framework with gene shuffling to optimize a FAR for the activity on acyl-ACP and improve fatty alcohol production.

Despite recent advances in high-throughput combinatorial mutagenesis assays, the number of labeled sequences available to predict molecular functions has remained small for the vastness of the sequence space combined with the ruggedness of many fitness functions. While deep neural networks (DNNs) can capture high-order epistatic interactions among the mutational sites, they tend to overfit to the small number of labeled sequences available for training. Here, we developed Epistatic Net (EN), a method for spectral regularization of DNNs that exploits evidence that epistatic interactions in many fitness functions are sparse. We built a scalable extension of EN, usable for larger sequences, which enables spectral regularization using fast sparse recovery algorithms informed by coding theory. Results on several biological landscapes show that EN consistently improves the prediction accuracy of DNNs and enables them to outperform competing models which assume other priors. EN estimates the higher-order epistatic interactions of DNNs trained on massive sequence spaces-a computational problem that otherwise takes years to solve., Finding a biologically-relevant inductive bias for training DNNs on large fitness landscapes is challenging. Here, the authors propose a method called Epistatic Net that improves DNN prediction accuracy and interpretation speed by integrating the knowledge that higher-order epistatic interactions are usually sparse.

The substantiation of an adaptive planting system for the natural frame of Voronezh is presented. A forest-park green belt has been identified, the components of which make up a single natural ecological system. It is indicated that the oak forests of the urban district of Voronezh occupy an area of 4500 hectares. Using the indicators of the normal head water level, average width, average depth, length of the coastline, regulation coefficient, the morphometric characteristics and standard head water levels of the Voronezh reservoir were determined. On the basis of functional zoning in the Voronezh reservoir, three zones are identified and their characteristics are given. It was identified that the forest cover of the Central Chernozem region is 9 %, which is lower than the normative indicators of forest cover for creating the natural frame in Voronezh. It is shown that the scientific objects of the Educational and Experimental Forestry Enterprise of the Voronezh State Forestry University (VGLTU) for the creation of forest cultures of the main forest-forming species in the Central Chernozem Region confirm the concept of creating adaptive landscaping systems for the natural frame of the city of Voronezh. 140 species in the genus of birch have been identified, which can be used for adaptive landscaping systems in the area under consideration. The stated description of the Great Voronezh Ecological Trail (BVET) along the entire route, characterized by the presence of unique natural landscape, botanical, geological, garden and park objects. An upward trend in the indicator of recreation of adaptive landscaping systems from 31,904.3 hectares was recorded up to 32421.2 hectares in 2020. It is noted that the climatic and natural conditions of this territory are favorable for growing an expanded range of tree species and shrubs, primarily the main forest-forming species.

Developing the wind energy industry (WEI) has been a long-term strategy in China for responding to the energy crisis and greenhouse gas emissions. However, China’s WEI faces various barriers despite its favorable prospects. This paper aims to propose a path analysis approach for exploring the best methods and timing for overcoming diverse barriers in developing WEI systematically, especially in terms of revealing in which phases these barriers should take priority. A multidimensional barrier set categorized by institutional, economic and financial, social, technical, and market factors is identified via the integration of a literature review and empirical interviews. A path analysis approach based on the grey group DEMATEL-NK model is constructed, where the causalities and intensities among barriers are extracted to design a path simulation algorithm with heuristics. The optimal path for overcoming barriers in developing the WEI is generated through a simulation of the search and optimization process used to climb to the fitness landscape peak. The results show that the different barriers to completing the legal and regulatory framework should be prioritized to reach adequate financial incentives and coordination among stakeholders, and that an efficient talent cultivation system should be developed and industry-academic cooperation should be strengthened to increase technical and R&D capabilities in the short term. The optimal path presents a strategic instrument for managers to use to better develop sustainable and clean wind energy systems from the novel insights of prioritizing to overcome barriers.

In recent years, many different differential evolution (DE) variants have been proposed to solve real-world optimization problems. However, the performance of them is largely determined by the selection of the mutation strategy, an approach to choose favorable mutation strategy when solving various optimization problems has attracted increasing attention recently. In this paper, we propose a DE with an adaptive mutation operator based on fitness landscape (FLDE). The application of fitness landscape to DE requires three stages. First, we analyzed the fitness landscape features of each benchmark training function, a total of 45 benchmark functions are taken from CEC2014 and 2015. Then, the relationship between three mutation strategies and fitness landscape features is trained by random forest (RF) offline. Finally, the trained RF is used to predict which mutation strategy should be utilized to perform mutation operator for each problem during the evolutionary process. Besides, a historical memory parameter adaption mechanism and population size linear reduction are applied to the FLDE. The CEC2017 benchmark set is utilized to perform the experiments, and five well-known DE variant algorithms are compared with the FLDE algorithm. The experimental results attest that the proposed FLDE algorithm is highly competitive with the other five DE algorithms.

NeuroEvolution (NE) refers to a family of methods for optimizing Artificial Neural Networks (ANNs) using Evolutionary Computation (EC) algorithms. NeuroEvolution of Augmenting Topologies (NEAT) is considered one of the most influential algorithms in the field. Eighteen years after its invention, a plethora of methods have been proposed that extend NEAT in different aspects. In this article, we present a systematic literature review (SLR) to list and categorize the methods succeeding NEAT. Our review protocol identified 232 papers by merging the findings of two major electronic databases. Applying criteria that determine the paper's relevance and assess its quality, resulted in 61 methods that are presented in this article. Our review article proposes a new categorization scheme of NEAT's successors into three clusters. NEAT-based methods are categorized based on 1) whether they consider issues specific to the search space or the fitness landscape, 2) whether they combine principles from NE and another domain, or 3) the particular properties of the evolved ANNs. The clustering supports researchers 1) understanding the current state of the art that will enable them, 2) exploring new research directions or 3) benchmarking their proposed method to the state of the art, if they are interested in comparing, and 4) positioning themselves in the domain or 5) selecting a method that is most appropriate for their problem.

Cancer treatment effectiveness could be improved if it were possible to accurately anticipate the response of the tumor to treatment. Writing in Nature, Salehi et al. combine single-cell genomics and mathematical modeling to measure cancer subclone fitness and use these measurements to accurately predict the future trajectory of cancer evolution.

In the light of recent developments, the use of biological techniques in crop cultivation technologies needs to be expanded. The role of crop rotations with a high proportion of legume grasses, green manure crops and intermediate crops is increasing. The purpose of the research is to study soil fertility dynamics in terms of the main nutritional elements and determine their balance for the development of the adaptive landscape farming system in the North-Eastern region of the European part of Russia. Experiments were carried out on the experimental plots of the FARC of the North-East in 2002–2009. Soil – sod-podzolic middle loamy. We analyzed data of a long-term stationary experiment in four grain-fallow-grass field crop rotations. Meteorological conditions during the years of research were close to the average long-term values and favorable for crop cultivation. The introduction of low doses of fertilizers (N45P45K45 as mineral, 11.5–22.5 t/ha as green manure) increased the content of mobile phosphorus in the soil by 11–14 mg/kg (LSD05 = 9.5) in crop rotations with occupied and green manure fallows. An increase in exchangeable potassium by 15 mg/kg (LSD05 = 14.1) was noted in crop rotation with two clover fields for green manure. A positive balance of nitrogen and phosphorus was obtained in crop rotation with occupied and green manure fallows. The nitrogen balance was with a higher intake of +157.8 – +483.5 kg/ha and balance intensity – 115–140 %. The balance of phosphorus has developed at +6.0…+49.6 kg/ha. The positive balance of potassium was only in crop rotations with the introduction of green manure fertilizers and intermediate crop (+69.9…+175.8 kg/ha). The intensity of the balance of these elements was 101–118 %. In the control crop rotation with bare fallow at a minimum intake of nutrients, the negative balance of nitrogen, phosphorus and potassium was obtained; the intensity was below 100 %. Productivity of crop rotation was at the level of 4.90–5.41 thousand fodder units. According to the optimal combination of productivity and the balance of nutritional elements, crop rotation with two clover fields for green manure and an intermediate crop stood out.

A major issue while conceiving or parameterizing an optimization heuristic is to ensure an appropriate balance between exploitation and exploration of the search. Evolution strategies and neighborhood-based metaheuristics constitute relevant high-level frameworks, which ease the problem solving but are often complex to configure. Moreover, their effective behavior, according to the particularities of the search landscapes, remains difficult to grasp. In this paper, we deeply investigate the sampled walk search algorithm, which is a local search equivalent of the $$(1,\lambda )$$ -evolution strategy, considering that the neighborhood relation describes mutation possibilities. We specifically designed experiments to better understand the behavior of such a strategy offering a fine way to deal with the exploration versus exploitation dilemma. The main contribution is the analysis of search trajectories by evaluating and visualizing both their width (exploration) and their height (exploitation). More generally, we aim at bringing insights about the behavior of the $$(1,\lambda )$$ -ES in a discrete optimization context and within a fitness landscape perspective.

International audience

Settlement distribution modelling assumes that human spatial behaviour can be revealed by using the correct set of variables. However, the regional variables of large-scale gradients are only considered rarely. Also, settlement modelling applications are often impeded by fragmentary and error-prone datasets. Thus, in current-day Lithuania, in order to study the dynamics of settlement distributional rules, we implemented a modelling strategy that involves the use of statistical and machine learning methods with null model simulations and that is less affected by the incompleteness of datasets and the scales of variables. In this study, we used regional variables in order to test their usefulness while providing a general understanding of the large-scale processes that governed settlement distribution in prehistory. This study encompassed six time intervals between the Palaeolithic and Iron Ages. The obtained results consisted of regional settlement predictive maps, estimations of random forest variable importance and performances of GAM models, as well as determinations of Besag's L-functions of settlement patterns and Euclidean distances between the regional variables at the settlement sites. We detected decreasing spatial autocorrelation of the settlement prediction maps and settlement patterns, declining variable importance and model performance, as well as increasing regional environmental diversity of the settlement sites. A marked manifestation of these trends was observed between the Neolithic and Bronze Ages. This also corresponded to a major change in the constitution of the most important variables of the settlement distributions. We associate these findings with technological and lifestyle changes in the southern Baltic region, interpreting them as causing a roughening of the human adaptive landscape – increasing the specificity, number and complexity of the settlement distributional rules. Our study revealed that, even for incomplete datasets and spatial autocorrelation, the application of advanced statistical/machine learning techniques was able to provide new insights into the processes that governed the socio-cultural evolution and spatial dynamics of humans in prehistory.

Understanding how evolutionary agents behave in complex environments is a challenging problem. Agents can be faced with complex fitness landscapes derived from multi-stage tasks, interaction with others, and limited environmental feedback. Agents that evolve to overcome these can sometimes access greater fitness, as a result of factors such as cooperation and tool use. However, it is often difficult to explain why evolutionary agents behave in certain ways, and what specific elements of the environment or task may influence the ability of evolution to find goal-achieving behaviours; even seemingly simple environments or tasks may contain features that affect agent evolution in unexpected ways. We explore principled simplification of evolutionary agent-based models, as a possible route to aiding their explainability. Using the River Crossing Task (RCT) as a case study, we draw on analysis in the Minimal River Crossing (RC-) Task testbed, which was designed to simplify the original task while keeping its key features. Using this method, we present new analysis concerning when agents evolve to successfully complete the RCT. We demonstrate that the RC- environment can be used to understand the effect that a cost to movement has on agent evolution, and that these findings can be generalised back to the original RCT. Then, we present new insight into the use of principled simplification in understanding evolutionary agents. We find evidence that behaviour dependent on features that survive simplification, such as problem structure, are amenable to prediction; while predicting behaviour dependent on features that are typically reduced in simplification, such as scale, can be invalid.

In recent years, codon substitution models based on the mutation–selection principle have been extended for the purpose of detecting signatures of adaptive evolution in protein-coding genes. However, the approaches used to date have either focused on detecting global signals of adaptive regimes—across the entire gene—or on contexts where experimentally derived, site-specific amino acid fitness profiles are available. Here, we present a Bayesian site-heterogeneous mutation–selection framework for site-specific detection of adaptive substitution regimes given a protein-coding DNA alignment. We offer implementations, briefly present simulation results, and apply the approach on a few real data sets. Our analyses suggest that the new approach shows greater sensitivity than traditional methods. However, more study is required to assess the impact of potential model violations on the method, and gain a greater empirical sense its behavior on a broader range of real data sets. We propose an outline of such a research program.

Precise engineering of biological systems requires quantitative, high-throughput measurements, exemplified by progress in directed evolution. New approaches allow high-throughput measurements of phenotypes and their corresponding genotypes. When integrated into directed evolution, these quantitative approaches enable the precise engineering of biological function. At the same time, the increasingly routine availability of large, high-quality data sets supports the integration of machine learning with directed evolution. Together, these advances herald striking capabilities for engineering biology.

Rail guide vehicle (RGV) problems have the characteristics of fast running, stable performance, and high automation. RGV dynamic scheduling has a great impact on the working efficiency of an entire automated warehouse. However, the relative intelligent optimization research of different workshop components for RGV dynamic scheduling problems are insufficient scheduling in the previous works. They appear idle when waiting, resulting in reduced operating efficiency during operation. This article proposes a new distance landscape strategy for the RGV dynamic scheduling problems. In order to solve the RGV dynamic scheduling problem more effectively, experiments are conducted based on the type of computer numerical controller (CNC) with two different procedures programming model in solving the RGV dynamic scheduling problems. The experiment results reveal that this new distance landscape strategy can provide promising results and solves the considered RGV dynamic scheduling problem effectively.

Wright’s “adaptive landscape” has been influential in evolutionary thinking but controversial, especially because the landscape that organisms encounter is altered by the evolutionary process itself and the effects organisms have on their environments. Lewontin offered a mathematical heuristic describing the coupling of niche construction and adaptive evolution. Here, we propose a “dual landscape” model to view these relationships. Our model represents change as simultaneous movement on two landscapes, each a function of phenotype and environment. This model clarifies the evolutionary feedback generated by niche construction. We relate our model to Lewontin’s niche construction equations and illustrate it with three examples.

The climatic, soil, ecological and economic features of the natural landscapes in the Krasnodar Krai were analyzed. The land structure of cultivated land in Western Ciscaucasia was given; the current state of soils in the Krasnodar Krai was analyzed. The changes in the content of humus in the arable layer over the century were considered. The qualitative state of agricultural lands, such as the susceptibility to erosion, the content of humus, and indicators of the nutrients balance in the soil was shown. A conclusion the existing zonal agriculture systems do not use the landscape diversity of the region was drawn. The need for switching to the adaptive landscape system of agriculture was substantiated. Using the GIS technologies, the boundaries of the natural landscapes in the Krasnodar Krai and the areas of agricultural land was determined by types and the degree of the manifestation of erosion processes. An eco-economic assessment of the field crop rotations adapted to the natural landscape by cultivated lands was made.

Today’s knowledge work environment is one of frequent interruptions, unnecessary distractions, essential interactions and valuable fragments of solitary work. There is an urgent need to better unde...

GIS project was developed for the radioactively contaminated territory of the Tulskii NII. It was created in order to project the adaptive landscape agricultures. It was based on the information on the concentrations of 137Cs radionuclide in soil, compiled over 16 years. Electronic maps have been developed to create a GIS project and included the location of agricultural lands; crop rotation systems; distribution of specific activity values for artificial 137Cs radionuclide in agricultural lands; agrochemical indexes (the humus content, potassium content, contribution of phosphorus, the acidity), soil types, relief. The created GIS project and the corresponding data bases will be used to collect, store and analyse the results of the survey in order to project the adaptive landscape agricultures. Key words: GIS project, adaptive landscape agriculture, agricultural lands, radiation safety.

Due to that the performance of differential evolution (DE) significantly depends on offspring generation strategies, various DE variants have been reported with the improved mutation operators. However, on the one hand, the mutation operators in most DE variants are guided by the elites in terms of the fitness value, without considering their distribution information in the fitness landscape. It may lead to the population be evolving towards the unpromising areas more frequently if these elites are clustered in a locally optimal region. On the other hand, in most DE variants, the evolutionary information of the potential trial vectors is not fully utilized to guide the search, which will hamper the local exploitation in the promising regions that they are located in. To overcome these weaknesses, this article proposes an enhanced DE framework (DELDG) with a leaders-detection-and-guidance mechanism that consists of an adaptive leaders detection (ALD) and a neighborhood-based tournament selection (NTS). With these two novel operators, DELDG can not only guide the mutation process of each individual with multiple promising leaders detected by ALD, but also accelerate the convergence speed with the competition among the potential trial vectors by NTS. Therefore, DELDG is characterized by the explicit detection of the promising leaders according to their fitness values and distribution information and the effective use of the potential trial vectors in the neighborhood of each leader. Compared with 36 excellent DE variants and evolutionary algorithms (EAs), the experimental results on 28 IEEE CEC2013 real-parameter functions and 17 IEEE CEC2011 real-world problems have demonstrated the competitive performance of DELDG.

Rodrigues, N. M., Silva, S., & Vanneschi, L. (2020). A Study of Generalization and Fitness Landscapes for Neuroevolution. IEEE Access, 8, 108216-108234. [9113453]. https://doi.org/10.1109/ACCESS.2020.3001505 Fitness landscapes are a useful concept for studying the dynamics of meta-heuristics. In the last two decades, they have been successfully used for estimating the optimization capabilities of different flavors of evolutionary algorithms, including genetic algorithms and genetic programming. However, so far they have not been used for studying the performance of machine learning algorithms on unseen data, and they have not been applied to studying neuroevolution landscapes. This paper fills these gaps by applying fitness landscapes to neuroevolution, and using this concept to infer useful information about the learning and generalization ability of the machine learning method. For this task, we use a grammar-based approach to generate convolutional neural networks, and we study the dynamics of three different mutations used to evolve them. To characterize fitness landscapes, we study autocorrelation, entropic measure of ruggedness, and fitness clouds. Also, we propose the use of two additional evaluation measures: density clouds and overfitting measure. The results show that these measures are appropriate for estimating both the learning and the generalization ability of the considered neuroevolution configurations. publishersversion published

Meta-heuristics perform differently depending on the problem instance they are solving, meaning that manually choosing an algorithm is not trivial and an automatic selection is desirable. This task can be addressed using a meta-learning approach, which relates the characteristics of the problem instances to the performance of a set of solving algorithms. Therefore, the success of such approach is based on the quality of the extracted set of features. Some studies have proposed the use of features based on Fitness Landscape Analysis (FLA) to characterize optimization problems. However, extracting measures based on FLA usually requires a high computational effort. In a previous work, we have employed meta-heuristic selection on the Quadratic Assignment Problem (QAP) using some FLA measures. This research extends our study by including additional FLA meta-features, by using a less costly extraction method, and by considering more QAP instances. In total, we built five multi-label datasets, each composed of meta-features that were extracted by different sampling sizes, and then we used them to train Random Forest classifiers. Besides presenting satisfactory classification performance, and a decrease in time consumption in relation to our previous work, this selection approach was able to achieve better solution costs over the set of QAP instances if compared to running the meta-heuristics individually.

Protein sequences contain rich information about protein evolution, fitness landscapes, and stability. Here we investigate how latent space models trained using variational auto-encoders can infer these properties from sequences. Using both simulated and real sequences, we show that the low dimensional latent space representation of sequences, calculated using the encoder model, captures both evolutionary and ancestral relationships between sequences. Together with experimental fitness data and Gaussian process regression, the latent space representation also enables learning the protein fitness landscape in a continuous low dimensional space. Moreover, the model is also useful in predicting protein mutational stability landscapes and quantifying the importance of stability in shaping protein evolution. Overall, we illustrate that the latent space models learned using variational auto-encoders provide a mechanism for exploration of the rich data contained in protein sequences regarding evolution, fitness and stability and hence are well-suited to help guide protein engineering efforts., Multiple sequence alignments of proteins carry information about evolution, the protein’s fitness landscape and its stability in the face of mutations. Here, the authors demonstrate the utility of latent space models learned using variational autoencoders to infer these properties from sequences.

Machine learning could enable an unprecedented level of control in protein engineering for therapeutic and industrial applications. Critical to its use in designing proteins with desired properties, machine learning models must capture the protein sequence-function relationship, often termed fitness landscape. Existing bench-marks like CASP or CAFA assess structure and function predictions of proteins, respectively, yet they do not target metrics relevant for protein engineering. In this work, we introduce Fitness Landscape Inference for Proteins (FLIP), a benchmark for function prediction to encourage rapid scoring of representation learning for protein engineering. Our curated tasks, baselines, and metrics probe model generalization in settings relevant for protein engineering, e.g. low-resource and extrapolative. Currently, FLIP encompasses experimental data across adeno-associated virus stability for gene therapy, protein domain B1 stability and immunoglobulin binding, and thermostability from multiple protein families. In order to enable ease of use and future expansion to new tasks, all data are presented in a standard format. FLIP scripts and data are freely accessible at https://benchmark.protein.properties.

Adeno-associated viruses (AAVs) hold tremendous promise as delivery vectors for clinical gene therapy, but they need improvement. AAVs with enhanced properties, such as more efficient and/or cell-type specific infection, can be engineered by creating a large, diverse starting library and screening for desired phenotypes, in some cases iteratively. Although this approach has succeeded in numerous specific cases, such as infecting cell types from the brain to the lung, the starting libraries often contain a high proportion of variants unable to assemble or package their genomes, a general prerequisite for engineering any gene delivery goal. Herein, we develop and showcase a machine learning (ML)-based method for systematically designing more effective starting libraries — ones that have broadly good packaging capabilities while being as diverse as possible. Such carefully designed but general libraries stand to significantly increase the chance of success in engineering any property of interest. Furthermore, we use this approach to design a clinically-relevant AAV peptide insertion library that achieves 5-fold higher packaging fitness than the state-of-the-art library, with negligible reduction in diversity. We demonstrate the general utility of this designed library on a downstream task to which our approach was agnostic: infection of primary human brain tissue. The ML-designed library had approximately 10-fold more successful variants than the current state-of-the-art library. Not only should our new library serve useful for any number of other engineering goals, but our library design approach itself can also be applied to other types of libraries for AAV and beyond.

Computational models of managerial search often build on backward-looking search based on hill-climbing algorithms. Regardless of its prevalence, there is some evidence that this family of algorithms does not universally represent managers’ search behavior. Against this background, the paper proposes an alternative algorithm that captures key elements of Simon’s concept of satisficing which received considerable support in behavioral experiments. The paper contrasts the satisficing-based algorithm to two variants of hill-climbing search in an agent-based model of a simple decision-making organization. The model builds on the framework of NK fitness landscapes which allows controlling for the complexity of the decision problem to be solved. The results suggest that the model’s behavior may remarkably differ depending on whether satisficing or hill-climbing serves as an algorithmic representation for decision-makers’ search. Moreover, with the satisficing algorithm, results indicate oscillating aspiration levels, even to the negative, and intense—and potentially destabilizing—search activities when intra-organizational complexity increases. Findings may shed some new light on prior computational models of decision-making in organizations and point to avenues for future research.

A 2D replica of real-world terrain was created in a physics simulation environment, allowing a robot to “imagine” a simulated version of itself navigating the terrain. The physics of the environment simulates the movement of robot parts and its interaction with objects, thus avoiding the need for explicitly programming various calculations. Since the complexity of motion increases with each degree of freedom of the robot’s joints, the utility of uniform randomness was also investigated as an alternative to computational intelligence algorithms for exploring the fitness landscape. Such techniques potentially simplify the algorithmic complexity of programming multi-jointed robots and could adapt by dynamically adjusting simulation parameters, on encountering environments with varied gravity, viscosity or traction.

This study builds on previous work in which Mask R-CNN was used to quantify common rust severity on maize leaves grown in a greenhouse environment. The Mask R-CNN outperformed standard image processing techniques when applied to the rust quantification task. One of the challenges with using Mask R-CNN is the large amount of parameters that need to be tuned, this study examines automated tuning of these parameters. This study is the first to perform hyper-parameter tuning on the Mask R-CNN at this scale as well as using a genetic algorithm (GA). The results gathered from the GA implementation is compared to that of the prior results from the Mask R-CNN with manually selected hyper-parameters. This work concludes that using a genetic algorithm to determine optimal hyper-parameter values for the Mask R-CNN is comparable to the current state of the art for this problem domain. Future work includes performing a fitness landscape analysis and potential development of new genetic operators.

Premature convergence can be detrimental to the performance of search methods, which is why many search algorithms include restart strategies to deal with it. While it is common to perturb the incumbent solution with diversification steps of various sizes with the hope that the search method will find a new basin of attraction leading to a better local optimum, it is usually unclear whether this strategy is effective. To establish a connection between restart effectiveness and properties of a problem, we introduce a new property of fitness landscapes termed Neighbours with Similar Fitness. We conjecture that this property is true for many PLS-complete problems, and we argue that the effectiveness of a restart strategy depends on this property.

This paper presents a surrogate-based approach that uses a relatively simple population-based optimisation algorithm, a basic Differential Evolution algorithm (DE), and experiments with two complementary approaches to construct a surrogate. This surrogate-based optimisation uses a predictive model in-line and decides whether to calculate a candidate individual (using the simulation model) or discard it as part of the optimisation process. The complementary approaches for the design of the surrogate are (1) a traditional regression-based surrogate that approximates the surface of the fitness landscape using a supervised continuous machine learning algorithm (XGBoost), and (2) a pairwise approach that models the surrogate as a binary classification problem for a machine learning algorithm (in this experiment we proposed a Decision Tree binary classifier). Although there is no statistical difference in the performance of both surrogate approaches, the surface/regression one obtains a slightly better performance when the execution is limited to 200 fitness evaluations. In contrast, the pairwise/classification approach obtains the lowest value and a lower mean in executions with 750 fitness evaluations.

The performance of an evolutionary algorithm on a particular optimization problem highly depends on the choice of the algorithm's parameters. Fitness landscape analysis can help learn important characteristics of the optimization problem that may influence the optimal parameter values. Using this knowledge for parameter choice can be an important step in designing fitness-aware algorithms. In this paper, we present an approach to automatic parameter choice that uses exploratory landscape analysis and machine learning. Using a neural network trained on a dataset of optimal parameter choices for particular landscape features of the W-model problem we choose parameters for the (1 + (λ, λ)) genetic algorithm. The experimental results suggest that the neural network is capable of providing good parameter choices based on the landscape features computed by the flacco package.

There are some natural similarities among fitness landscapes of different modals involved in a multimodal optimization problem (MMOP). However, existing multimodal evolutionary algorithms (MMEAs) tend to handle these modals separately, which limits their performance to a great extent. As the first attempt, this study proposes a knowledge transfer-based MMEA by mining and exploiting modal similarities. To this end, a center-aligned normalization strategy (CANS) is first designed to map all the subpopulations corresponding to identified modals into a unified 0-1 space, and then a distribution similarity-based transfer strategy (DSTS) is proposed to guide knowledge transfer among subpopulations. DSTS explicitly quantifies the similarity between two subpopulations and probabilistically transfers elite individuals in the subpopulation of the largest similarity to the target subpopulation. By integrating CANS and DSTS into a well-known MMEA, i.e., NEA2, the final knowledge transfer-based NEA named KTNEA is developed. Experimental results on CEC’2013 niching benchmark suite indicate that KTNEA performs competitively in comparison with state-of-the-art MMEAs.

Effective implementations of Memetic Algorithms often integrate, within their design, problem-based pieces of information. When no information is known, an efficient MA can still be designed after a preliminary analysis of the problem. This approach is usually referred to as Fitness Landscape Analysis (FLA). This paper proposes a FLA technique to analyse the epistasis of continuous optimisation problems and estimate those directions, within a multi-dimensional space, associated with maximum and minimum directional derivatives. This estimation is achieved by making use of the covariance matrix associated with a distribution of points whose objective function value is below (in case of minimisation) a threshold. The eigenvectors and eigenvalues of the covariance matrix provide important pieces of information about the geometry of the problem and are then used to design a memetic operator that is a local search belonging to the family of generalised Pattern Search. A restarting mechanism enables a progressive characterisation of the fitness landscape. Numerical results show that the proposed approach successfully explore ill-conditioned basins of attractions and outperforms the standard pattern search as well as a pattern search recently proposed in the literature and partially based on a similar design logic. The proposed local search based on FLA also displays a performance competitive with that of other types of local search.

In this paper, we propose an automated method for the development of new representations of sequences. For this purpose, we introduce a two-way mapping from variable length sequence representations to a latent representation modelled as the bottleneck of an LSTM (long short-term memory) autoencoder. Desirable properties of such mappings include smooth fitness landscapes for optimization problems and better evolvability. This work explores the capabilities of such latent encodings in the context of optimization of 3D structures. Various improvements are adopted that include manipulating the autoencoder architecture and its training procedure. The results of evolutionary algorithms that use different variants of automatically developed encodings are compared.

In local search algorithms, the pivoting rule determines which neighboring solution to select and thus strongly influences the behavior of the algorithm and its capacity to sample good-quality local optima. The classical pivoting rules are first and best improvement, with alternative rules such as worst improvement and maximum expansion recently studied on hill-climbing algorithms. This article conducts a thorough empirical comparison of five pivoting rules (best, first, worst, approximated worst and maximum expansion) on two benchmark combinatorial problems, NK landscapes and the unconstrained binary quadratic problem (UBQP), with varied sizes and ruggedness. We present both a performance analysis of the alternative pivoting rules within an iterated local search (ILS) framework and a fitness landscape analysis and visualization using local optima networks. Our results reveal that the performance of the pivoting rules within an ILS framework may differ from their performance as single climbers and that worst improvement and maximum expansion can outperform classical pivoting rules.

Debugging is a time-consuming task for software engineers. Automated Program Repair (APR) has proved successful in automatically fixing bugs for many real-world applications. Search-based APR generates program variants that are then evaluated on the test suite of the original program, using a fitness function. In the vast majority of search-based APR work only the Boolean test case result is taken into account when evaluating the fitness of a program variant. We pose that more fine-grained fitness functions could lead to a more diverse fitness landscape, and thus provide better guidance for the APR search algorithms. We thus present 2Phase, a fitness function that also incorporates the output of test case failures, and compare it with ARJAe, that shares the same principles, and the standard fitness, that only takes the Boolean test case result into consideration. We conduct the comparison on 16 buggy programs from the QuixBugs benchmark using the Gin genetic improvement framework. The results show no significant difference in the performance of all three fitness functions considered. However, Gin was able to find 8 correct fixes, more than any of the APR tools in the recent QuixBugs study.

One of the issues general AI game players are required to deal with is the different reward systems in the variety of games they are expected to be able to play at a high level. Some games may present plentiful rewards which the agents can use to guide their search for the best solution, whereas others feature sparse reward landscapes that provide little information to the agents. The work presented in this paper focuses on the latter case, which most agents struggle with. Thus, modifications are proposed for two algorithms, Monte Carlo Tree Search and Rolling Horizon Evolutionary Algorithms, aiming at improving performance in this type of games while maintaining overall win rate across those where rewards are plentiful. Results show that longer rollouts and individual lengths, either fixed or responsive to changes in fitness landscape features, lead to a boost of performance in the games during testing without being detrimental to non-sparse reward scenarios.

Purpose Organizations increasingly operate under volatile, uncertain, complex and ambiguous (VUCA) conditions. Traditional command-and-control leadership can be ineffective in such chaotic environments. The purpose of this paper is to outline an alternative model to help leaders and organizations navigate effectively through VUCA environments. By developing three fundamental capacities (absorptive, adaptive and generative), leaders can cultivate organizations capable of continuous synchronization with their fitness landscapes. Central tenets of the framework include diversity, slack, learning, humility, reflection in action and abductive logic. Design/methodology/approach This framework is designed based on literature insights, conceptual analysis and experts’ judgment. The paper integrates knowledge from a variety of disciplines and interprets them through the lens of complex adaptive systems. Findings This paper argues for a process centered, contemplative approach to organizational leadership and development. By providing the underlying rationale for the proposed interventions (e.g. Ashby’s law of requisite variety), the paper also reorients busy leaders’ mental models to show why these time investments are worth implementing. Practical implications This actionable framework can help leaders and organizations be more effective operating in a VUCA context. Originality/value This paper provides a historic context as to why prediction and certainty are favored leadership strategies, why these approaches are no longer suitable and specific steps leaders can take to develop absorptive, adaptive and generative capacities to transform their organizations. Its scholarly contribution is the synthesis of disparate bodies of literature, weaving those multiple academic perspectives into a practical roadmap to enhance organizational leadership.

The no-wait flow shop scheduling problem (NWFSP) is one of the essential models in the manufacturing systems. In this paper, the fitness landscape of the factorial representation for NWFSP with the makespan criterion is studied. The encoding and decoding schemes based on the factorial representation are constructed to transfer the permutation domain to the integer domain. The position-type distributions and fitness distance correlation are implemented to analyze the fitness landscape of the classic benchmarks. The multiple big valleys' structure in the fitness landscape is confirmed through the observation of fitness distance plot and the analysis of factorial coding theory. The various local optima and high ruggedness of the fitness landscape are visualized through the statistical results of position-type distributions. The results of fitness landscape analysis show the suitability of the landscape for the searchability with evolutionary algorithms and local search methods for solving NWFSP.

Fitness landscape analysis has been effectively used to analyze the characteristics of combinatorial optimization problems (COPs) while investigating the behavior of applied algorithms. However, most COPs are high dimensional and an intuitive understanding of traditional fitness landscape analysis is challenging. To address this issue, the present study proposes a spatial-domain fitness landscape analysis framework. This framework aims to visualize the fitness landscapes regarding of a target COP and evaluate the properties of the COP. To construct the landscapes, a mapping strategy inspired by lexicographic order is designed to establish the connection between the high-dimensional and low-dimensional spaces. Accordingly, a spatial-domain parameter, namely, the slope of the digital elevation models (DEMs), is utilized to assess the ruggedness of landscapes. As an auxiliary value, the neutrality ratio is introduced to reflect the degree of neutrality. In addition, a parameter control mechanism based on these landscape metrics is advanced to enhance the adaptability of the algorithm. Finally, the experimental results reveal that the spatial-domain fitness landscape can help extract the characteristics of COPs and study the behavior of algorithms effectively. In addition, the proposed adaptive parameter control strategy significantly improves the performance of the algorithm for solving COPs.

Recently, the improvements of artificial bee colony (ABCs) have attracted increasing interest in the studies of single optimization problems. However, most existing work of ABC aims to design new solution search equations and is still challenged when solving optimization problems with variable linkages. To overcome this limit, an adaptive encoding learning for ABCs (AEL + ABCs) is proposed in this paper. In AEL + ABCs, the solution search equations are encoded in both natural coordinate system and eigen coordinate system guided by covariance matrix learning. The purpose of the former is to maintain the diversity of population, while the latter aims at directing the evolution of population toward the promising directions by identifying the properties of fitness landscape. In addition, an adaptive selection mechanism is used to achieve a good tradeoff between convergence and diversity. For the comparison purposes, the proposed AEL strategy is applied to eight ABCs and their performance is tested on 30 CEC2014 benchmark functions. Experiment results show that the proposed AEL + ABCs can significantly improve the performance of the state-of-the-art ABCs in the majority of the benchmark functions.

Dynamic fitness landscape analyses contain different metrics to attempt to analyze optimization problems. In this article, some of dynamic fitness landscape metrics are selected to discuss differential evolution (DE) algorithm properties and performance. Based on traditional differential evolution algorithm, benchmark functions and dynamic fitness landscape measures such as fitness distance correlation for calculating the distance to the nearest global optimum, ruggedness based on entropy, dynamic severity for estimating dynamic properties, a fitness cloud for getting a visual rendering of evolvability and a gradient for analyzing micro changes of benchmark functions in differential evolution algorithm, the authors obtain useful results and try to apply effective data, figures and graphs to analyze the performance differential evolution algorithm and make conclusions. Those metrics have great value and more details as DE performance.

International audience

Background: Recently, there has been a growing interest in applying immune checkpoint blockers (ICBs), so far used to treat cancer, to patients with bacterial sepsis. We aimed to develop a method for predicting the personal benefit of potential treatments for sepsis, and to apply it to therapy by meropenem, an antibiotic drug, and nivolumab, a programmed cell death-1 (PD-1) pathway inhibitor.Methods: We defined an optimization problem as a concise framework of treatment aims and formulated a fitness function for grading sepsis treatments according to their success in accomplishing the pre-defined aims. We developed a mathematical model for the interactions between the pathogen, the cellular immune system and the drugs, whose simulations under diverse combined meropenem and nivolumab schedules, and calculation of the fitness function for each schedule served to plot the fitness landscapes for each set of treatments and personal patient parameters.Results: Results show that treatment by meropenem and nivolumab has maximum benefit if the interval between the onset of the two drugs does not exceed a dose-dependent threshold, beyond which the benefit drops sharply. However, a second nivolumab application, within 7–10 days after the first, can extinguish a pathogen which the first nivolumab application failed to remove. The utility of increasing nivolumab total dose above 6 mg/kg is contingent on the patient's personal immune attributes, notably, the reinvigoration rate of exhausted CTLs and the overall suppression rates of functional CTLs. A baseline pathogen load, higher than 5,000 CFU/μL, precludes successful nivolumab and meropenem combination therapy, whereas when the initial load is lower than 3,000 CFU/μL, meropenem monotherapy suffices for removing the pathogen.Discussion: Our study shows that early administration of nivolumab, 6 mg/kg, in combination with antibiotics, can alleviate bacterial sepsis in cases where antibiotics alone are insufficient and the initial pathogen load is not too high. The study pinpoints the role of precision medicine in sepsis, suggesting that personalized therapy by ICBs can improve pathogen elimination and dampen immunosuppression. Our results highlight the importance in using reliable markers for classifying patients according to their predicted response and provides a valuable tool in personalizing the drug regimens for patients with sepsis.

This chapter presents surrogate-assisted evolutionary algorithms for single-objective optimization that employ multiple surrogates. Multiple surrogates can not only improve the prediction performance and estimate the degree of prediction uncertainty, but also capture both global and local features of the fitness landscape. The multiple surrogates can be used as an ensemble, in parallel, hierarchically, or in an interleaving way. Finally, we describe a method for adaptively selecting one surrogate at a particular search stage from a pool of surrogates according to their performance in the history.

Boolean functions have numerous applications in domains as diverse as coding theory, cryptography, and telecommunications. Heuristics play an important role in the construction of Boolean functions with the desired properties for a specific purpose. However, there are only sparse results trying to understand the problem’s difficulty. With this work, we aim to address this issue. We conduct a fitness landscape analysis based on Local Optima Networks (LONs) and investigate the influence of different optimization criteria and variation operators. We observe that the naive fitness formulation results in the largest networks of local optima with disconnected components. Also, the combination of variation operators can both increase or decrease the network size. Most importantly, we observe correlations of local optima’s fitness, their degrees of interconnection, and the sizes of the respective basins of attraction. This can be exploited to restart algorithms dynamically and influence the degree of perturbation of the current best solution when restarting.

Multimodality plays a key role as one of the most challenging problem characteristics in the common understanding of solving optimization tasks. Based on insights from the single-objective optimization domain, local optima are considered to be (deceptive) traps for optimization approaches such as gradient descent or different kinds of neighborhood search. Consequently, as continuous multi-objective (MO) problems usually combine multiple (often multimodal) single-objective problems, multimodality is considered an important challenge for MO problems as well. In fact, even very simple MO problems possess a multimodal landscape due to the interactions among its objectives. Thus, modern benchmarks name multimodality an important problem characteristic, while at the same time, heuristic algorithms (like evolutionary algorithms) are expected to be almost mandatory for handling multimodality in an effective way. Here, we continue our previous work by (1) formally defining multimodality in the continuous MO setting, (2) provide techniques for visualizing landscapes of continuous MO problems—not only in the objective but also in the decision space—to improve the intuition regarding continuous MO multimodality, and (3) analyze MO problems based on examples from an extensive test-bed. Thereby, we provide the tools for displaying and detecting basins of attraction, as well as superpositions of local optima, in the decision space of the landscapes. Most important, and maybe unexpected, we are able to show that multimodality in continuous MO optimization differs largely from our understanding of multimodality in the single-objective domain: for simple MO optimization approaches, local efficient sets are often no traps. Instead, locality can even be exploited to slide toward the global efficient set.

This chapter examines the fundamental limits placed on algorithm capabilities due to the internal solution representation used by a search algorithm, and its implications for mode discovery and problem tractability. It presents various approaches to visualizing the problem landscape, based on both exhaustive enumeration and sampling—which may be of use to both practitioners and problem owners. It also provides the first comprehensive examination of the widely used IEEE CEC 2013 benchmark multimodal problems using local optima networks. These visualize the fitness landscape as a directed graph, and convey such information as local optima quality, basin size, and how easy it is to traverse the fitness landscape from one local optimum to another.