Exploitation Strategies Research Articles

Mobile robot path planning using deep deterministic policy gradient with differential gaming (DDPG-DG) exploration

Mobile robot path planning involves decision-making in uncertain, dynamic conditions, where Reinforcement Learning (RL) algorithms excel in generating safe and optimal paths. The Deep Deterministic Policy Gradient (DDPG) is an RL technique focused on mobile robot navigation. RL algorithms must balance exploitation and exploration to enable effective learning. The balance between these actions directly impacts learning efficiency.This research proposes a method combining the DDPG strategy for exploitation with the Differential Gaming (DG) strategy for exploration. The DG algorithm ensures the mobile robot always reaches its target without collisions, thereby adding positive learning episodes to the memory buffer. An epsilon-greedy strategy determines whether to explore or exploit. When exploration is chosen, the DG algorithm is employed. The combination of DG strategy with DDPG facilitates faster learning by increasing the number of successful episodes and reducing the number of failure episodes in the experience buffer. The DDPG algorithm supports continuous state and action spaces, resulting in smoother, non-jerky movements and improved control over the turns when navigating obstacles. Reward shaping considers finer details, ensuring even small advantages in each iteration contribute to learning.Through diverse test scenarios, it is demonstrated that DG exploration, compared to random exploration, results in an average increase of 389% in successful target reaches and a 39% decrease in collisions. Additionally, DG exploration shows a 69% improvement in the number of episodes where convergence is achieved within a maximum of 2000 steps.

Role of Archaeomalacological Research in Sri Lanka: A Synthesis

This article focuses on the role of malacological research in Sri Lanka and the significance it holds within the field of archaeology. We aim to explore and synthesize the diverse applications of malacology to understand past societies and environments that once thrived on the island. Malacology, the study of Mollusca, is an interdisciplinary field that combines elements of biology, ecology, and archaeology. In the last fifty years, malacological research has gained traction within archaeological investigations, contributing valuable insights into human-environment interactions, subsistence patterns, and cultural dynamics. By integrating data from archaeological excavations with molluscan remains found at sites, researchers can reconstruct past environments, identify resource exploitation strategies, and shed light on human dietary choices. We explore the implications of malacological studies for understanding trade networks, migration patterns, and the socio-economic complexities of ancient Sri Lankan societies. We summarize existing literature on archaeomalacological research worldwide, as well as specific case studies conducted within Sri Lanka. By placing these findings within the broader context of Sri Lanka’s archaeology, this research aims to bridge existing knowledge gaps and contribute to a more comprehensive understanding of the island’s rich cultural heritage, highlighting the pivotal role of archaeomalacology and its potential to enhance interpretations of past societies and environments. This synthesis seeks to examine the potential contributions of archaeomalacology in the understanding of Sri Lanka’s past.

Forms of Khaya exploitation and conservation strategies on the Atacora range in Togo

Les espèces de Khaya font l’objet de surexploitation entrainant la menace de leur disparition. Cette étude sur les formes d’exploitation et stratégies de conservation des pieds de Khaya a été réalisée sur la chaîne de l’Atacora au Togo. Elle a pour objectif général de contribuer à la gestion durable des Khaya spp. au Togo. Elle vise spécifiquement à : i) présenter l’état des pieds de Khaya spp. ; (ii) évaluer l’ampleur des me-naces anthropiques sur les pieds de Khaya et (iii) localiser les semenciers de Khaya spp. La collecte des données forestières s’est basée sur un échantillonnage orienté par la présence des Khaya spp. sur des placettes de 2500 m² dans les champs et jachères, de 900 m² dans les savanes, de 400 m² dans les forêts claires et agroforêts et de 500 m² dans les formations ripicoles. Les paramètres dendrométriques sont relevés pour les tiges d’arbres. Le nombre des pieds de Khaya coupés et des semenciers ont été compté et le code sanitaire des arbres de Khaya relevé. Les résultats montrent que les pieds de Khaya ont une densité de 29,67 pieds/ha avec une surface terrière de 13,36 m2/ha dans la zone écologique II contre une densité de 27,91 pieds/ha et une surface terrière de 5,96 m2/ha dans la zone IV. Les formes d’exploitation sur les pieds de Khaya sont les coupes de bois, les écorçages, les élagages, les prélèvements de racines et la récolte des fruits. Les écorçages (90,90 %) et les coupes de bois (68,83 %) sont plus fréquents que les autres formes d’exploitation. Des possibilités de restaurations des Khaya exis-tent dans la zone d’étude avec la présence des semenciers de Khaya senegalensis dans la zone II et de Khaya grandifoliola dans la zone IV. Il urge d’adopter des mesures de gestion durable de Khaya spp. en vue de limiter la dégradation de leurs peuplements naturels.

Automatic differentiation of Eucalyptus species through Sentinel-2 images, Worldview-3 images and LiDAR data

Eucalyptus constitutes one of the most common tree genera used in forest plantations worldwide. In Europe, Eucalyptus trees are especially common in the Northwest of the Iberian Peninsula, E. nitens and E. globulus being the most commonly cultivated species. Each species presents particularities that lend to different exploitation strategies and industrial usages. Therefore, updated knowledge about the abundance and spatial distribution of the different species is important for forest planning. This is a special challenge for areas where forest land is highly fragmented. Remote sensing has been used to efficiently monitor the distribution of the Eucalyptus genera, however little research has been able to map specific Eucalyptus species. This study evaluates the efficiency of Sentinel-2 data, Worldview-3 images, and Airborne LiDAR data in the differentiation of E. nitens and E. globulus. Supervised classifications were performed using neural networks for these data sets both individually and in combination. The highest accuracies were obtained when using Sentinel-2 data in combination with LiDAR point clouds and when using Sentinel-2 data in a multitemporal approach. The best time of year to differentiate between the two species is during the emergence of spring shoots. Worldview-3 images have a moderate capacity to differentiate between the two species, although this is increased when textural metrics are included. This study can serve as the basis for generating Eucalyptus species distribution maps, which will allow for improved forest management and planning.

Human and birds

The paper presents the study of avifauna from the hunter-gatherer sites at the Dnieper-Dvina basin spanning time period from the 6th to 3rd millennia BC. A total of 669 bird bones were identified and attributed to 46 different bird taxa, representing resident and migrant birds. They belong to four habitat groups: waterfowl, forest, woodside and meadow-steppe. The dominance of waterfowl birds follows the common strategy of aquatic resources exploitation. Changes in the procurement strategies, use and symbolic meanings of birds can be envisaged. Reconstructed regional mean temperature fluctuations suggest a particular influence on breeding biology and migration patterns of different species.

World Hyper-Heuristic: A novel reinforcement learning approach for dynamic exploration and exploitation

In the real world, there are many complex problems in engineering. Every problem has a level of computational complexity, starting from simple problems and reaching NP-hard problems. Np-hard problems do not have a definite answer. Therefore, hyper-heuristic algorithms try to optimize NP-hard problems. Hyper-heuristics optimize complex search problems using a combination plate of exploration and exploitation strategies. The current algorithms need more generalizability, handling specific data types, limitations to a particular search problem, and weak performance. We propose a World hyper-heuristic (World) to address the issues using a novel reinforcement learning method. World, in two steps of rewarding and selection, dynamically switches between exploration and exploitation strategies provided in an infinite pool of meta-heuristics. We evaluated the performance of our proposed method in three phases. First, we optimized the standard functions as artificial Np-hard problems. Then, we compared real engineering examples related to discrete Np-hard problems. Finally, we implemented and analyzed problems with the continuous Np-hard problems. Our extensive comparisons with the state-of-the-art algorithms demonstrate the World’s outperformance in handling varied search problems with any data type. Among all the findings, World finds the shortest path of length 4.33e + 05, far shorter than the results of the state-of-the-art work, in benchmarked data of 10,000 real cities.

중국 내 한국 중소기업의 동적 역량이 성과에 미치는 영향

The purpose of this study is to analyze the relationship between SMEs’ dynamic capabilities and performance in China and find a moderating effect of exploration and exploitation strategy on this relationship. With 200 samples, we found that the networking capability, internationalization capability, and management capability of Korean SMEs in China contribute to improving the company’s performance. Furthermore, regarding the relationship between dynamic capabilities and performance, the moderating effect of exploitation strategy was observed only in the positive relationship between management capability and performance. This indicates that in China, where motivations such as cost reduction and local market penetration are strong, efficient management leading to cost reduction is the most effective strategy. On the other hand, the moderating effect of exploration strategy on the relationship between networking capability and performance, as well as the moderating effect of exploitation strategy on the relationship between internationalization capability and performance, was not supported.

Neurophysiological mechanisms of exploration and exploitation in high-functioning autism: Magnetoencephalographic study

Intolerance of uncertainty and high sensitivity to the threat of failure are believed to contribute to chronic anxiety in individuals with high-functioning autism. This study examines the influence of these personality traits on the brain processes involved in decision-making in a probabilistic setting among autism patients. Furthermore, we explore the causal relationship between these personality traits and chronic anxiety in individuals with high-functioning autism. Twenty-one high-functioning autistic individuals who are intolerant to uncertainty, aged 19 to 46, and 21 neurotypical volunteers of the same age, carried out a probabilistic choice experiment with two alternatives. One option yielded a monetary profit of 70% of the time, while the other brought only a 30% gain. Following each selection, the subjects received feedback on their decision and gradually learned, through trial and error, to prefer the more advantageous option. From this point forward, we consider frequent selections of a beneficial stimulus to align with the internal utility model, specifically the exploitation strategy. On the other hand, sporadic choices of a detrimental stimulus correspond to an exploration strategy, which may prove disadvantageous in a stable environment but enables adaptation to unforeseen changes in the surroundings. We hypothesized that there are characteristic differences in brain activity reflecting exploration and exploitation strategies between groups, which would emerge during the decision-making period and after internal feedback evaluation regarding advantageous and disadvantageous choices [1]. Beta oscillations (16–30 Hz) were analyzed in the magnetoencephalographic recordings. The decrease in beta oscillation power below baseline during the period between stimulus presentation and subjects’ response indicates activation in brain regions related to decision-making strategy. Conversely, an increase in beta oscillation power following feedback signaling a disadvantageous choice reflects a mechanism that reinforces internal value models in current task conditions [1]. The cortical sources of beta oscillations in 448 cortical areas were estimated using the sLoreta technique at the single-trial level. Mixed linear models (LMM) were used for statistical analysis with correction for multiple comparisons via the FDR method for the number of cortical areas analyzed. The study concentrated on the time interval for decision-making (–900 to –300 ms prior to the motor action of choice) and the post-feedback period (500 to 900 ms after the commencement of the feedback presentation) [1]. Based on questionnaire results, individuals with high-functioning autism exhibited a significantly lower tolerance for uncertainty and a greater intolerance for uncertainty when compared to neurotypical participants. The study generated two primary findings. Initially, the extent of brain activity during decision-making differed in control subjects and subjects with autism based on the type of choice, with exact opposition. Opting for an advantageous vs. disadvantageous choice was linked with reduced activation of the inferior temporal, parietal, and medial frontal cortex regions in control subjects and elevated activation in these regions for subjects with autism spectrum disorder. These findings suggest that neurotypical individuals utilize fewer neural resources and exhibit decreased emotional response while deciding in favor of a known profitable outcome compared to opting for an uncertain choice — one that is more likely to result in a negative outcome based on previous experiences. Individuals with autism allocate disproportionate amounts of attention and emotional resources towards planning actions that are deemed safe and offer only a probable (though not definite) advantage. Conversely, the prospect of failure due to risky behavior induces a comparatively muted response in their brains. Second, our internal evaluation of feedback focused on variations in functional activity within the orbitofrontal and lateral prefrontal cortical areas during exploratory (disadvantageous) choices between individuals with autism and control subjects. As previously reported, neurotypical subjects demonstrated significant beta synchronization following negative feedback after disadvantageous choices [1]. Contrary to control participants, individuals with autism spectrum disorder exhibited a lack of synchronization in frontal beta oscillations following losses incurred from unfavorable selections. This observation may indicate insufficient reinforcement of the internal utility model, which typically strengthens in response to negative outcomes that align with predicted results [1]. Overall, our study found that individuals on the autism spectrum with a high intolerance for uncertainty exhibit significantly increased activation of brain decision-making systems in situations that are perceived to be low-risk and with a high probability of a successful outcome. These findings clarify the reasons behind the puzzling rise in anxiety and autonomic reactivity among individuals in situations where they anticipate rewards, which are not inherently aversive unlike fear of punishment [2].

Genetic diversity and divergence among native and translocated populations of the golden flathead goby Glossogobius aureus (Gobiiformes: Gobiidae) in Philippine lakes

The golden flathead goby Glossogobius aureus is a native species in the Philippines, Australia, Japan, Taiwan, and many other countries in Asia. In the Philippines, it is an important food fish as it is commonly caught in major lakes. In this study, a total of 307 specimens morphologically identified as G. aureus were sampled from nine major lakes in the Philippines and were sequenced for their mitochondrial cytochrome b (cyt b) gene. Two hundred sixty of the 307 cyt b sequences had sequence similarities of ≥ 99% with G. aureus reference sequence in GenBank, while the remaining 47 (all from Lake Lanao) had sequence similarities of only 95% and were thus designated as Glossogobius cf. aureus and treated as a separate population. The sequences were then analyzed to examine the pattern of genetic diversity, relatedness, divergence, and demographic history among native and translocated populations of the species. Twenty-nine haplotypes were recovered, of which four haplotypes were shared among three to seven populations. Only one haplotype each was found in the native population in Lake Buhi and translocated population in Lake Paoay. Low haplotype and low nucleotide diversities were found for the populations in Laguna de Bay, Lanao, Bato, Buhi, Paoay, and Sebu lakes, which indicate founder event for the introduced populations in Lanao, Paoay, and Sebu lakes and recent genetic bottleneck for the native populations in Laguna de Bay, Bato, and Buhi. In contrast, high haplotype but low nucleotide diversities were found for the native populations of Taal, Naujan, and Buluan lakes, signifying a recent bottleneck followed by population expansion. Pairwise FST values showed generally large (FST = 0.168–0.249) to very large (FST = 0.302–1.000) genetic divergence between populations except between Laguna de Bay and Lake Bato, Laguna de Bay and Lake Buhi, and Lake Bato and Lake Buhi populations, which showed nonsignificant genetic differentiation. Lake Buluan and Lake Sebu populations showed moderate genetic differentiation (FST = 0.098). Neutrality tests showed significant negative Tajima’s D and Fu’s FS values only for the population from Laguna de Bay, which suggests that the population is undergoing expansion. These results are important for establishing scientifically sound strategies for effective conservation and sustainable exploitation of G. aureus in the Philippines.

Enhancing Semantic Segmentation through Reinforced Active Learning: Combating Dataset Imbalances and Bolstering Annotation Efficiency

This research addresses the challenges of training large semantic segmentation models for image analysis,focusing on expediting the annotation process and mitigating imbalanced datasets. In the context of imbalanceddatasets, biases related to age and gender in clinical contexts and skewed representation in natural images can affectmodel performance. Strategies to mitigate these biases are explored to enhance efficiency and accuracy in semanticsegmentation analysis. An in-depth exploration of various reinforced active learning methodologies for imagesegmentation is conducted, optimizing precision and efficiency across diverse domains. The proposed frameworkintegrates Dueling Deep Q-Networks (DQN), Prioritized Experience Replay, Noisy Networks, and EmphasizingRecent Experience. Extensive experimentation and evaluation of diverse datasets reveal both improvements andlimitations associated with various approaches in terms of overall accuracy and efficiency. This research contributesto the expansion of reinforced active learning methodologies for image segmentation, paving the way for moresophisticated and precise segmentation algorithms across diverse domains. The findings emphasize the need fora careful balance between exploration and exploitation strategies in reinforcement learning for effective imagesegmentation.

Review: Toward sustainable management of groundwater in the deserts of Egypt

To achieve its ambitious plans to reclaim its deserts through mega projects, Egypt is heavily relying on fossil or little-recharged groundwater. This article revisits the results and methodologies of the studies conducted over the last two decades on groundwater management and uses in the Western Desert. Most previous studies aimed at simulating different groundwater abstraction scenarios by modeling local areas in aquifer systems, but with poor definitions of boundary conditions and limited historical data. Studies were constrained by the unavailability of data, access difficulties, and high collection costs in desert lands. Thus, to propose reliable sustainable groundwater resources development plans and recommendations for future protection strategies, an open-access monitoring network representing regional aquifers is needed. More investigations based on extensive field visits are essential to monitor environmental, economic, and social conditions, identify constraints, and learn lessons for reclaiming desert lands. Moreover, this review highlighted the need to frame a rational strategy for the long-term sustainable exploitation of non-renewable groundwater in the aquifer systems of Egypt and develop an appropriate exit strategy for desert communities in case of serious water resource depletion.

SHFuzz: Service handler-aware fuzzing for detecting multi-type vulnerabilities in embedded devices

Embedded devices in IoT are of great convenience to our daily lives and industries, but they also introduce multi-type vulnerabilities. Most vulnerabilities reside in various handlers of service program. However, existing fuzzing methods existing fuzzing methods can neither efficiently nor effectively to discover these vulnerabilities: they can not extract the service handler alone and properly pool resources for testing them.In this study, we propose a novel service handlers aware fuzzing method to efficiently discover multi-type vulnerabilities in embedded devices. Our key observation is that service program dispatches the service handlers (the code snippet which implements the desired work of user request) in two general ways. Meanwhile, multi-type vulnerabilities usually happen at sensitive APIs during the execution of service handler. So in our method, we first design a static analysis method to extract the info of service handlers and sensitive APIs. Then during fuzzing, the service handler coverage strategy guides the fuzzer to cover more service handlers; after the sensitive APIs are reached, the sensitive APIs exploitation strategy is proposed to strengthen the ability and efficiency of discovering vulnerabilities. Based on our method, we implement a prototype system, named SHFuzz. Experiments on 19 target programs from popular brand devices show that, SHFuzz outperforms the state-of-art tool FIRM-AFL in the coverage of sensitive APIs, the diversity of detected vulnerabilities and efficiency. Furthermore, SHFuzz discovers 16 zero-day vulnerabilities and reports them to vendors.

Diversification of faunal exploitation strategy and human-climate interaction in Southern China and Southeast Asia during the last deglaciation

Southern China and Southeast Asia were favourable habitats for foragers during the Last Glacial Maximum (∼25–18 ka BP) and the Last Deglaciation (∼18–11.7 ka BP), despite various climate fluctuation. However, the underlying subsistence strategies in these areas remains unclear, due to the lack of systematic chronological and archaeological data. The current study reports new results of radiocarbon dating, U-series dating, and zooarchaeological analysis from the Naminan cave site, located at the boundary between Southern China and Southeast Asia. It reveals that deer was consumed as the primary source of meat (91.7 %), supplemented by turtles, wild oxen, monkeys, wild boars, and mustelids during ∼17.35–16.49 ka BP. The subsequent period of ∼15.55–15.35 ka BP saw a radical declining proportion of deer, with various new species being consumed, such as shrews, porcupines, fish, crab, canidae, and wild cat. Human occupation of the Naminan cave site might have lasted until the Early Holocene. Comparing archaeological and paleoclimate evidence from Southern China and Southeast Asia, it is reasonable to argue that a progressively drier and colder climate during ∼17.35–15.35 ka BP appeared to trigger a remarkable decline in human activity but greater range of diets in locations such as the Naminan cave and Yuchanyan sites, facilitating further expansion of hunting-gathering groups during the late phase of the Last Deglaciation with warm and humid climate. This study contributes fresh data and perspective to a greater understanding of the evolution of human-climate interactions in low-latitude regions during Late Paleolithic period.

Efficient property-oriented optimization of magnetic high-entropy metallic glasses via a multi-stage design strategy

High-entropy metallic glasses (HE-MGs) have drawn much attention as promising multifunctional materials combining the excellent soft magnetic properties of traditional metallic glasses and impressive mechanical properties, thermal stability, corrosion resistance, etc., of solid solution high entropy alloys. Property optimization of HE-MGs is of great significance for promoting their engineering applications. However, various constituent elements and the high chemical complexity make the possible alloying composition space extremely massive, which is very challenging for the rational design of HE-MGs. In this work, we proposed a multi-stage optimization strategy based on machine learning (ML) to accelerate the rational design of magnetic HE-MGs with desired properties. The huge composition search space was significantly narrowed by the ML-based phase prediction model and constraints from user preferences. Utility functions based on the exploitation and exploration strategy were designed to find the global optimization solutions, i.e., alloying compositions. Experiments were conducted as concept validation, and new Fe-Co-Ni-Si-B HE-MGs with balanced saturation magnetic flux density and mixing entropy were developed.

TAICHI: Transform Your Secret Exploits Into Mine From a Victim's Perspective

Acquiring and analyzing exploits, which take advantage of vulnerabilities to conduct malicious actions, are crucial for victims (and defenders) when responding to system compromising incidents. However, exploits are sensitive and valuable assets that are not available to victims. The most common resource available for victims to investigate is network traffic, which covers the exploitation period. Thus reconstructing exploits from network traffic is demanded. In practice, the reconstruction process is performed manually, thus inefficient and non-scalable. In this paper, we present an automated solution <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">TAICHI</monospace> to reconstruct exploits from network traffic, able to generate replica exploits and facilitate timely incident analysis. By nature, a working exploit has to satisfy (1) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">path constraints</i> which ensure the program path same as the original exploit’s is explored and the same vulnerability is triggered, and (2) <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">exploit constraints</i> which ensure the same exploitation strategy is applied, e.g., to bypass deployed defenses or to stitch multiple gadgets together. We propose a hybrid solution to this problem by integrating techniques including multi-version execution (MVE), dynamic taint analysis (DTA), and concolic execution. We have implemented a prototype of <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">TAICHI</monospace> on x86 and x86-64 Linux and tested it on the Cyber Grand Challenge (CGC) dataset, several Capture the Flag (CTF) challenges, and Metasploit exploit modules targeting real world applications. The evaluation results showed that <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">TAICHI</monospace> could reconstruct exploits efficiently with a high success rate. Moreover, it could be applied to production environments without disrupting running services, and could reconstruct exploits even if only one round of exploitation traffic is available.

Understanding patch foraging strategies across development.

Patch foraging is a near-ubiquitous behaviour across the animal kingdom and characterises many decision-making domains encountered by humans. We review how a disposition to explore in adolescence may reflect the evolutionary conditions under which hunter-gatherers foraged for resources. We propose that neurocomputational mechanisms responsible for reward processing, learning, and cognitive control facilitate the transition from exploratory strategies in adolescence to exploitative strategies in adulthood - where individuals capitalise on known resources. This developmental transition may be disrupted by psychopathology, as there is emerging evidence of biases in explore/exploit choices in mental health problems. Explore/exploit choices may be an informative marker for mental health across development and future research should consider this feature of decision-making as a target for clinical intervention.

EXISTENCE OF MAXIMUM OF TIME AVERAGED HARVESTING IN THE KPP-MODEL ON SPHERE WITH PERMANENT AND IMPULSE COLLECTION

On a two-dimensional sphere, a distributed renewable resource is considered, the dynamics of which is described by a model of the Kolmogorov–Petrovsky–Piskunov–Fisher type, and the exploitation of this resource, carried out by constant or periodic impulse harvesting. It is shown that after choosing an admissible exploitation strategy, the dynamics of the resource tend to the limiting dynamics corresponding to this strategy, and that there is an admissible harvesting strategy that maximizes the time averaged harvesting of the resource.

Market Orientation's Neglected Role in Marketing Exploitation and Exploration Tradeoffs

The goal of marketing exploitation methods, sometimes referred to as marketing exploration, can be to increase the company's knowledge and capabilities or to improve its current capabilities. According to a study on organisational strategy and learning, using both strategies can compromise each company's operational processes and harm its financial success. Market organisation, in our opinion, facilitates the flow of market information between the two strategy processes, integrates the two strategies by acting as a dynamic market, and aids businesses in integrating marketing exploitation and exploration strategies. Using complementary high-level marketing exploration and marketing exploitation strategies, companies with strong market orientation can improve the financial performance of new products, as measured at two different points in time, according to a study of the packaged food industry in Indonesia. However, as the trade-off foresees, businesses with a poor market orientation are unable to benefit from these techniques.

Groundwater quality modeling and determining critical points: a comparison of machine learning to Best-Worst Method.

In Iran, similar to other developing countries, groundwater quality has been seriously threatened. Therefore, this study aimed to apply Machine Learning Algorithms (MLAs) in Groundwater Quality Modeling (GQM) and determine the optimal algorithm using the Best-Worst Method (BWM) in Ardabil province, Iran. Groundwater quality parameters included calcium (Ca2+), magnesium (Mg2+), sodium (Na+), potassium (K+), chlorine (Cl-), sulfate (SO4-), total dissolved solids (TDS), bicarbonate (HCO3-), electrical conductivity (EC), and acidity (pH). In the following, seven MLAs, including Support Vector Regression (SVR), Random Forest (RF), Decision Tree Regressor (DTR), K-Nearest Neighbor (KNN), Naïve Bayes, Simple Linear Regression (SLR), and Support Vector Machine (SVM), were used in the Python programming language, and groundwater quality was modeled. Finally, BWM was used to validate the results of MLAs. The results of examining the error statistics in determining the optimal algorithm in groundwater quality modeling showed that the RF algorithm with values of MAE = 0.28, MSE = 0.12, RMSE = 0.35, and AUC = 0.93 was selected as the most optimal MLA. The Schoeller diagram also showed that various ion ratios, including Na+K, Ca2+, Mg2+, Cl-, and HCO3+CO3, in most of the sampled points had upward average values. Based on the results of the BWM method, it can be concluded that a great similarity was observed between the results of the RF algorithm and the classification of the BWM method. These results showed that more than 50% of the studied area had low quality based on hydro-chemical parameters of groundwater quality. The findings of this research can assist managers and planners in developing suitable management models and implementing appropriate strategies for the optimal exploitation of groundwater resources.

Surrogate Ensemble-Assisted Hyper-Heuristic Algorithm for Expensive Optimization Problems

This paper proposes a novel surrogate ensemble-assisted hyper-heuristic algorithm (SEA-HHA) to solve expensive optimization problems (EOPs). A representative HHA consists of two parts: the low-level and the high-level components. In the low-level component, we regard the surrogate-assisted technique as a type of search strategy and design the four search strategy archives: exploration strategy archive, exploitation strategy archive, surrogate-assisted estimation archive, and mutation strategy archive as low-level heuristics (LLHs), each archive contains one or more search strategies. Once the surrogate-assisted estimation archive is activated to generate the offspring individual, SEA-HHA first selects the dataset for model construction from three principles: All Data, Recent Data, and Neighbor, which correspond to the global and the local surrogate model, respectively. Then, the dataset is randomly divided into training and validation data, and the most accurate model built by polynomial regression (PR), support vector regression (SVR), and Gaussian process regression (GPR) cooperates with the infill sampling criterion is employed for solution estimation. In the high-level component, we design a random selection function based on the pre-defined probabilities to manipulate a set of LLHs. In numerical experiments, we compare SEA-HHA with six optimization techniques on 5-D, 10-D, and 30-D CEC2013 benchmark functions and three engineering optimization problems with only 1000 fitness evaluation times (FEs). The experimental and statistical results show that our proposed SEA-HHA has broad prospects for dealing with EOPs.