An interpretable machine learning framework for modeling and analysis of swarm robot behaviors

Mapping benthic sediment types and composition in a turbid Jamaican bay using hydroacoustic data and different spatially explicit interpolation techniques

The rapid population growth and continuous urban expansion in Bandung have contributed to volatile and escalating housing prices, creating significant challenges for market transparency and affordability. This study aims to develop and evaluate machine-learning models to predict house prices in the Bandung region using a publicly available dataset consisting of 7,609 property records. Following the CRISP-DM methodology, the research includes data exploration, preprocessing (outlier handling using IQR, one-hot encoding, and feature standardization), model training, and performance evaluation. Two regression models K-Nearest Neighbors (KNN) Regressor and Random Forest (RF) Regressor—were compared through systematic hyperparameter tuning using Grid Search and Random Search techniques. The experimental results show that the Random Forest Regressor achieves the best performance with an R² score of 0.7838 and a mean absolute error (MAE) of approximately Rp 399.7 million, outperforming the optimized KNN model. Feature importance analysis also indicates that land area, building area, and location are the most influential predictors of property prices. The findings highlight the effectiveness of ensemble methods in handling complex real-estate data and demonstrate the potential of machine-learning-based predictive tools to support buyers, sellers, and policymakers in making informed and data-driven decisions in the Bandung housing market.

Downscaling land surface temperatures at regional scales with random forest regression

The analysis of athletic performance has always aroused great interest from sport scientist. This study utilized machine learning methods to build predictive models using a comprehensive CrossFit (CF) dataset, aiming to reveal valuable insights into the factors influencing performance and emerging trends. Random forest (RF) and multiple linear regression (MLR) were employed to predict performance in four key weightlifting exercises within CF: clean and jerk, snatch, back squat, and deadlift. Performance was evaluated using R-squared (R2) values and mean squared error (MSE). Feature importance analysis was conducted using RF, XGBoost, and AdaBoost models. The RF model excelled in deadlift performance prediction (R2=0.80), while the MLR model demonstrated remarkable accuracy in clean and jerk (R2=0.93). Across exercises, clean and jerk consistently emerged as a crucial predictor. The feature importance analysis revealed intricate relationships among exercises, with gender significantly impacting deadlift performance. This research advances our understanding of performance prediction in CF through machine learning techniques. It provides actionable insights for practitioners, optimize performance, and demonstrates the potential for future advancements in data-driven sports analytics.

Abstract. Firn density plays a crucial role in assessing the surface mass balance of the Antarctic ice sheet. However, our understanding of the spatial and temporal variations in firn density is limited due to (i) spatial and temporal limitations of in situ measurements, (ii) potential modelling uncertainties, and (iii) lack of firn density products driven by satellite remote-sensing data. To address this gap, this paper explores the potential of satellite microwave radiometer (Special Sensor Microwave Imager/Sounder (SSMIS)) and scatterometer (Advanced Scatterometer (ASCAT)) observations for assessing spatial and temporal dynamics of dry-firn density over the Antarctic ice sheet. Our analysis demonstrates a clear relation between density anomalies at a depth of 40 cm and fluctuations in satellite observations. However, a linear relationship with individual satellite observations is insufficient to explain the spatial and temporal variation in snow density. Hence, we investigate the potential of a non-linear random forest (RF) machine learning approach trained on radiometer and scatterometer data to derive the spatial and temporal variations in dry-firn density. In the estimation process, 10 years of SSMIS observations (brightness temperature) and ASCAT observations (backscatter intensity) is used as input features to a random forest (RF) regressor. The regressor is first trained on time series of modelled density and satellite observations at randomly sampled pixels and then applied to estimate densities in dry-firn areas across Antarctica. The RF results reveal a strong agreement between the spatial patterns estimated by the RF regressor and the modelled densities. The estimated densities exhibit an error of ±10 kg m−3 in the interior of the ice sheet and ±35 kg m−3 towards the ocean. However, the temporal patterns show some discrepancies, as the RF regressor tends to overestimate summer densities, except for high-elevation regions in East Antarctica and specific areas in West Antarctica. These errors may be attributed to underestimations of short-term or seasonal variations in the modelled density and the limitations of RF in extrapolating values outside the training data. Overall, our study presents a potential method for estimating unknown Antarctic firn densities using known densities and satellite parameters.

Background: Modeling patient flow is crucial in understanding resource demand and prioritization. We study patient outflow from an open ward in an Australian hospital, where currently bed allocation is carried out by a manager relying on past experiences and looking at demand. Automatic methods that provide a reasonable estimate of total next-day discharges can aid in efficient bed management. The challenges in building such methods lie in dealing with large amounts of discharge noise introduced by the nonlinear nature of hospital procedures, and the nonavailability of real-time clinical information in wards.ObjectiveOur study investigates different models to forecast the total number of next-day discharges from an open ward having no real-time clinical data.MethodsWe compared 5 popular regression algorithms to model total next-day discharges: (1) autoregressive integrated moving average (ARIMA), (2) the autoregressive moving average with exogenous variables (ARMAX), (3) k-nearest neighbor regression, (4) random forest regression, and (5) support vector regression. Although the autoregressive integrated moving average model relied on past 3-month discharges, nearest neighbor forecasting used median of similar discharges in the past in estimating next-day discharge. In addition, the ARMAX model used the day of the week and number of patients currently in ward as exogenous variables. For the random forest and support vector regression models, we designed a predictor set of 20 patient features and 88 ward-level features.ResultsOur data consisted of 12,141 patient visits over 1826 days. Forecasting quality was measured using mean forecast error, mean absolute error, symmetric mean absolute percentage error, and root mean square error. When compared with a moving average prediction model, all 5 models demonstrated superior performance with the random forests achieving 22.7% improvement in mean absolute error, for all days in the year 2014.ConclusionsIn the absence of clinical information, our study recommends using patient-level and ward-level data in predicting next-day discharges. Random forest and support vector regression models are able to use all available features from such data, resulting in superior performance over traditional autoregressive methods. An intelligent estimate of available beds in wards plays a crucial role in relieving access block in emergency departments.

Quantitative structure-activity relationship (QSAR) regression models are mathematical ones which relate the structural properties of chemicals to the potencies of the biological activities of the chemicals. In QSAR models, the physical and chemical information of the molecules is encoded into quantitative numbers called descriptors. Recently, experimental test results (profiles) have been used as descriptors of chemicals. Profile QSAR 2.0 (pQSAR) model suggested by Martin et al., is a multitask, two step machine learning prediction method with a combination of random forest regressions (RFRs) and partial least squares regression (PLSR). In pQSAR model, one fills the profile table's missing values with RFRs and then builds PLSR using the profile predictions. Note that in the second step of the pQSAR method, PLSR's predictor variables are profiles; so activity values, and the response variables are also activity values. Thus we can use the PLSRs to update the profile table and then repeat the second step. In this work, we propose an extended model of pQSAR generated by RFRs and PLSRs. Experiment of updating the given full initially predicted profile table by two kinds of prediction models, RFRs and PLSRs, has been conducted iteratively for the PKIS and ChEMBL data sets. Even though prediction performance of individual combination of RFRs and PLSRs varies, the average of the all possible predicted profile tables for given iteration shows better performance. This ensemble model has better prediction performance in sense of Pearson's R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> compared to that of the pQSAR model.

In swarm robotics multiple robots collectively solve problems by forming advantageous structures and behaviors similar to the ones observed in natural systems, such as swarms of bees, birds, or fish. However, the step to industrial applications has not yet been made successfully. Literature is light on real-world swarm applications that apply actual swarm algorithms. Typically, only parts of swarm algorithms are used which we refer to as basic swarm behaviors. In this paper we collect and categorize these behaviors into spatial organization, navigation, decision making, and miscellaneous. This taxonomy is then applied to categorize a number of existing swarm robotic applications from research and industrial domains. Along with the classification, we give a comprehensive overview of research platforms that can be used for testing and evaluating swarm behavior, systems that are already on the market, and projects that target a specific market. Results from this survey show that swarm robotic applications are still rare today. Many industrial projects still rely on centralized control, and even though a solution with multiple robots is employed, the principal idea of swarm robotics of distributed decision making is neglected. We identified mainly following reasons: First of all, swarm behavior emerging from local interactions is hard to predict and a proof of its eligibility for applications in an industrial context is difficult to provide. Second, current communication architectures often do not match requirements for swarm communication, which often leads to a system with a centralized communication infrastructure. Finally, testing swarms for real industrial applications is an issue, since deployment in a productive environment is typically too risky and simulations of a target system may not be sufficiently accurate. In contrast, the research platforms present a means for transforming swarm robotics solutions from theory to prototype industrial systems.

Multi-scenario adaptive electronic nose for the detection of environmental odor pollutants.

Pollution is a major public health and human rights issue that disproportionately affects the poor and disadvantaged. However, academic studies have not fully explored the relationship between poverty and the spatial distribution of environmental pollution, notably in Thailand. Thus, this chapter aims to fill a void in academic research by exploring the implementation of machine learning in the estimation of poverty by training input data from widely available and accessible open source, including environmental pollution and road density. The poverty rate is obtained from the TPMAP website and then clustered into two groups, “high and low,” using hot spot analysis. The Google Earth Engine is used to extract pollution indexes such as carbon monoxide, formaldehyde, nitrogen, and sulfur dioxide, while road density is downloaded from OpenStreetMap. This study compares four different machine learning models: XGboost, lasso, random forest, and ridge regression for poverty estimation. The result of the study reveals that poverty-related areas are highly correlated with environmental pollution. The random forest technique has the best performance prediction among the four methods, with an R2 of 0.79. Finally, feature importance analysis is used to determine the most influential features in order to assist decision-makers in gaining a better understanding of poverty.KeywordsPollutionPovertyMachine learningGoogle Earth EngineGeospatial regression

This study investigated the feature importance of near-infrared spectra from random forest regression models constructed to predict the carbonization characteristics of hydrochars produced by hydrothermal carbonization of kraft lignin. The model achieved high coefficients of determination of 0.989, 0.988, and 0.985 with root mean square errors of 0.254, 0.003, and 0.008 when predicting the carbon content, atomic O/C ratio, and H/C ratio, respectively. The random forest models outperformed the multilayer perceptron models for all predictions. In the feature importance analysis, the spectral regions at 1600–1800 nm, the first overtone of C–H stretching vibrations, and 2000–2300 nm, the combination bands, were highly important for predicting the carbon content and O/C predictions, whereas the region at 1250–1711 nm contributed to predicting H/C. The random forest models trained with the high-importance regions achieved better prediction performances than those trained with the entire spectral range, demonstrating the usefulness of the feature importance yielded by the random forest and the feasibility of selective application of the spectral data.

This paper presents novel QSAR models for the prediction of antitrypanosomal activity among thiazolidines and related heterocycles. The performance of four machine learning algorithms: Random Forest regression, Stochastic gradient boosting, Multivariate adaptive regression splines and Gaussian processes regression have been studied in order to reach better levels of predictivity. The results for Random Forest and Gaussian processes regression are comparable and outperform other studied methods. The preliminary descriptor selection with Boruta method improved the outcome of machine learning methods. The two novel QSAR-models developed with Random Forest and Gaussian processes regression algorithms have good predictive ability, which was proved by the external evaluation of the test set with corresponding Q2ext =0.812 and Q2ext =0.830. The obtained models can be used further for in silico screening of virtual libraries in the same chemical domain in order to find new antitrypanosomal agents. Thorough analysis of descriptors influence in the QSAR models and interpretation of their chemical meaning allows to highlight a number of structure-activity relationships. The presence of phenyl rings with electron-withdrawing atoms or groups in para-position, increased number of aromatic rings, high branching but short chains, high HOMO energy, and the introduction of 1-substituted 2-indolyl fragment into the molecular structure have been recognized as trypanocidal activity prerequisites.

As global supply chains continue to expand in complexity and scale, the integration of intelligent, data-driven solutions within logistics operations becomes increasingly essential. Ports are evolving from traditional gateways into dynamic, technology-enabled hubs that demand advanced predictive capabilities and operational flexibility. With Porto Romano poised to become a major logistics and energy platform for Albania, the early adoption of machine learning (ML) tools is a strategic advantage. This study explores the application of two widely recognized ML algorithms- Random Forest Regression and Support Vector Machine (SVM) Regression- to estimate intermodal transport costs based on critical factors such as distance, travel time, cargo weight, and transport mode. A dataset comprising 250 transport cases was constructed to evaluate model performance using metrics such as Mean Squared Error (MSE), R² Score, Accuracy, and F1-Score. The data for the railway sector were simulated to support future port development, while the maritime and road sector data were derived from actual observations. Results indicate that Random Forest consistently outperformed SVM in predictive accuracy and robustness. Feature importance analysis further highlighted distance and travel time as the primary drivers of transport cost variations, aligning with operational expectations. By integrating ML-based cost prediction tools, Porto Romano could significantly improve its logistical efficiency, cost-effectiveness, and strategic competitiveness in the Balkan region. Moreover, this research underscores the broader value of machine learning in optimizing global intermodal logistics networks, offering pathways towards more sustainable and resilient supply chains. Future work will aim to refine these models with live operational data and expand into deep learning methodologies for even greater predictive capabilities

Swarm robotics is an approach to collective robotics that takes inspiration from the self-organized behaviors of social animals. Through simple rules and local interactions, swarm robotics aims at designing robust, scalable, and flexible collective behaviors for the coordination of large numbers of robots [1]. With the simplicity in individual robot design and flexibility in the dimension of their collectivity, swarm robot systems have received considerable attention in recent research. However, lots of previous literature has focused on collective behaviors of swarm robot systems to achieve higher capability in both communication and coordination, whereas, individuality of a specific swarm robot has been seldom addressed. Traditionally, because the design of an individual swarm robot is rather simple, using many robots is the only way to tackle complicated tasks. Considering expanding demands of robotic tasks with higher complexity, higher dimensionality, and different information density, other than employing the conventional low cost-effective approach of increasing the number of swarm, this project intends to adopt higher swarm individuality to obtain higher flexibility and re-configurability of the system. The proposed swarm design has been tested with a self-developed multi-target following system, the Auto-Cart (a new supermarket service system). In order to optimize the system performance using a simple individual robot design and a minimum number of swarm, a new navigation and communication algorithm was proposed. Simulation and experiment results verified the effectiveness and efficiency of the proposed system and algorithm.

This project presents a swarming and herding behaviour using simple robots. The main goal is to demonstrate the applicability of artificial intelligence (AI) in simple robotics that can then be scaled to industrial and consumer markets to further the ability of automation. AI can be achieved in many different ways; this paper explores the possible platforms on which to build a simple AI robots from consumer grade microcontrollers. Emphasis on simplicity is the main focus of this paper. Cheap and 8 bit microcontrollers were used as the brain of each robot in a decentralized swarm environment were each robot is autonomous but still a part of the whole. These simple robots don’t communicate directly with each other. They will utilize simple IR sensors to sense each other and simple limit switches to sense other obstacles in their environment. Their main objective is to assemble at certain location after initial start from random locations, and after converging they would move as a single unit without collisions. Using readily available microcontrollers and simple circuit design, semiconsistent swarming behaviour was achieved. These robots don’t follow a set path but will react dynamically to different scenarios, guided by their simple AI algorithm.