Explainable Machine Learning Research Articles

Abnormal phenotypic defects detection of jujube using explainable machine learning enhanced computer vision.

Jujube is susceptible to biotic and abiotic adversity stresses resulting in abnormal phenotypic defects. Therefore, abnormal phenotype fruits should be removed during postharvest sorting to increase added value. An improved maximum horizontal diameter linear regression (MHD-LR) method for size grading of jujube prior to detection of abnormal phenotypic defects was developed. The accuracy of the MHD-LR model is 95%, with an error of only 0.95mm. In addition, a method for detecting abnormal phenotypic defects in jujube was established. It can effectively and accurately classify seven kinds of jujube phenotypes (regular, irregular, wrinkled, moldy, hole-broken, skin-broken, and scarred). The data augmentation method based on linear interpolation can effectively expand the dataset with a variance of only 0.0006. Support vector machine-decision tree (SVMDT), logistic regression, back propagation neural network, and long short-term memory network models were established to classify jujube samples with different phenotypes, with accuracies of 99.57%, 99.00%, 99.14%, and 99.29%, respectively. The results showed that the SVMDT model had higher accuracy and explainability. This research is expected to provide a new method to improve the precise classification of abnormal phenotypic defects in postharvest jujube.

Predicting special forces dropout via explainable machine learning.

Selecting the right individuals for a sports team, organization, or military unit has a large influence on the achievements of the organization. However, the approaches commonly used for selection are either not reporting predictive performance or not explainable (i.e., black box models). In the present study, we introduce a novel approach to selection research, using various machine learning models. We examined 274 special forces recruits, of whom 196 dropped out, who performed a set of physical and psychological tests. On this data, we compared four machine learning models on their predictive performance, explainability, and stability. We found that a stable rule-based (SIRUS) model was most suitable for classifying dropouts from the special forces selection program. With an averaged area under the curve score of 0.70, this model had good predictive performance, while remaining explainable and stable. Furthermore, we found that both physical and psychological variables were related to dropout. More specifically, a higher score on the 2800m time, need for connectedness, and skin folds was most strongly associated with dropping out. We discuss how researchers and practitioners can benefit from these insights in sport and performance contexts.

Development and validation of an explainable machine learning model for predicting multidimensional frailty in hospitalized patients with cirrhosis.

We sought to develop and validate a machine learning (ML) model for predicting multidimensional frailty based on clinical and laboratory data. Moreover, an explainable ML model utilizing SHapley Additive exPlanations (SHAP) was constructed. This study enrolled 622 patients hospitalized due to decompensating episodes at a tertiary hospital. The cohort data were randomly divided into training and test sets. External validation was carried out using 131 patients from other tertiary hospitals. The frail phenotype was defined according to a self-reported questionnaire (Frailty Index). The area under the receiver operating characteristics curve was adopted to compare the performance of five ML models. The importance of the features and interpretation of the ML models were determined using the SHAP method. The proportions of cirrhotic patients with nonfrail and frail phenotypes in combined training and test sets were 87.8% and 12.2%, respectively, while they were 88.5% and 11.5% in the external validation dataset. Five ML algorithms were used, and the random forest (RF) model exhibited substantially predictive performance. Regarding the external validation, the RF algorithm outperformed other ML models. Moreover, the SHAP method demonstrated that neutrophil-to-lymphocyte ratio, age, lymphocyte-to-monocyte ratio, ascites, and albumin served as the most important predictors for frailty. At the patient level, the SHAP force plot and decision plot exhibited a clinically meaningful explanation of the RF algorithm. We constructed an ML model (RF) providing accurate prediction of frail phenotype in decompensated cirrhosis. The explainability and generalizability may foster clinicians to understand contributors to this physiologically vulnerable situation and tailor interventions.

Exploring the effects of 2D/3D building factors on urban energy consumption using explainable machine learning

The impact of various factors on building energy consumption (BEC) has garnered significant attention, as understanding these factors can help stakeholders optimize urban buildings' energy efficiency. However, most studies analyze these factors from a two-dimensional (2D) perspective (e.g., building materials, floor area ratio), overlooking the importance of emerging three-dimensional (3D) factors (e.g., building shading coefficient, building shape coefficient). This study aims to bridge this research gap by proposing a comprehensive framework to explore the impact of 2D/3D factors on urban BEC. The framework is validated using data from 818,199 buildings in New York City and employs an explainable machine learning method to reveal the contribution of 2D/3D factors to BEC. The results show that the importance of 2D and 3D factors for BEC is approximately 70 % and 30 %, respectively. Furthermore, this study considers the effects of building shade-related factors on BEC, finding their influence to be around 5–10 %. Based on the explainable artificial intelligence (AI) method results, a numerical dependency analysis is conducted for high-impact factors, providing decision-makers with a clearer understanding of how 2D/3D factors affect BEC. The novelty of this research lies in its comprehensive approach to investigating the impact of both 2D and 3D factors on urban BEC using advanced AI techniques. The findings offer valuable insights for urban building design and planning, particularly in the early stages, to promote energy efficiency. This study contributes to the growing body of knowledge on sustainable urban development and the application of advanced data analytics in building engineering.

Explainable machine learning and online calculators to predict heart failure mortality in intensive care units

Explainable machine learning and online calculators to predict heart failure mortality in intensive care units

Identifying Key Factors for Securing a Champions League Position in French Ligue 1 Using Explainable Machine Learning Techniques

Introduction: Performance analysis is essential for coaches and a topic of extensive research. The advancement of technology and Artificial Intelligence (AI) techniques has revolutionized sports analytics. Aim: The primary aim of this article is to present a robust, explainable machine learning (ML) model that identifies the key factors that contribute to securing one of the top three positions in the standings of the French Ligue 1, ensuring participation in the UEFA Champions League for the following season. Materials and Methods: This retrospective observational study analyzed data from all 380 matches of the 2022–23 French Ligue 1 season. The data were obtained from the publicly-accessed website “whoscored” and included 34 performance indicators. This study employed Sequential Forward Feature Selection (SFFS) and various ML algorithms, including XGBoost, Support Vector Machine (SVM), and Logistic Regression (LR), to create a robust, explainable model. The SHAP (SHapley Additive Explanations) model was used to enhance model interpretability. Results: The K-means Cluster Analysis categorized teams into groups (TOP TEAMS, 3 teams/REST TEAMS, 17 teams), and the ML models provided significant insights into the factors influencing league standings. The LR classifier was the best-performing classifier, achieving an accuracy of 75.13%, a recall of 76.32%, an F1-score of 48.03%, and a precision of 35.17%. “SHORT PASSES” and “THROUGH BALLS” were features found to positively influence the model’s predictions, while “TACKLES ATTEMPTED” and “LONG BALLS” had a negative impact. Conclusions: Our model provided satisfactory predictive accuracy and clear interpretability of results, which gave useful information to stakeholders. Specifically, our model suggests adopting a strategy during the ball possession phase that relies on short passes (avoiding long ones) and aiming to enter the attacking third and the opponent’s penalty area with through balls.

Investigating the contribution of image time series observations to cauliflower harvest-readiness prediction.

Cauliflower cultivation is subject to high-quality control criteria during sales, which underlines the importance of accurate harvest timing. Using time series data for plant phenotyping can provide insights into the dynamic development of cauliflower and allow more accurate predictions of when the crop is ready for harvest than single-time observations. However, data acquisition on a daily or weekly basis is resource-intensive, making selection of acquisition days highly important. We investigate which data acquisition days and development stages positively affect the model accuracy to get insights into prediction-relevant observation days and aid future data acquisition planning. We analyze harvest-readiness using the cauliflower image time series of the GrowliFlower dataset. We use an adjusted ResNet18 classification model, including positional encoding of the data acquisition dates to add implicit information about development. The explainable machine learning approach GroupSHAP analyzes time points' contributions. Time points with the lowest mean absolute contribution are excluded from the time series to determine their effect on model accuracy. Using image time series rather than single time points, we achieve an increase in accuracy of 4%. GroupSHAP allows the selection of time points that positively affect the model accuracy. By using seven selected time points instead of all 11 ones, the accuracy improves by an additional 4%, resulting in an overall accuracy of 89.3%. The selection of time points may therefore lead to a reduction in data collection in the future.

An explainable machine learning approach to predict the compressive strength of graphene oxide-based concrete

Graphene oxide (GO) has shown promise in improving concrete strength. Despite its frequent use in cement composites, its effect on concrete properties is less explored. The influence of GO on concrete remains unclear due to various interactions among constituents such as GO type and percentage, Superplasticiser (SP) type and percentage, dispersion technique, and curing age. Therefore, making prediction of compressive strength using simple mathematical formation is challenging. This study represents the first-ever attempt at developing a comprehensively validated machine learning model to predict GO-concrete's compressive strength characteristics by considering all key variable parameters. A comprehensive laboratory experiment program was conducted to collect the data required for training the machine learning (ML) models. Once the ML models were trained, they were used to predict the relationship of each of those input variables towards the compressive strength properties. Four machine learning models—(a) multiple linear regression (MLR), (b) k-nearest neighbour (KNN), (c) random forest (RF) and (d) extreme gradient boost (XGB)—were utilised to model the relationship between input parameters and compressive strength. Also, explainable machine learning (XML) methods were employed to elucidate the impact of mixed constituents on the compressive strength of GO concrete. The results from test predictions showcase that the XGB has attained a coefficient of determination (R2) of 0.981, coefficient of correlation (R) of 0.99, mean square error (MSE) of 0.9, mean normalised bias (MNB) of 0.004, scatter index (SI) of 0.2 with mean absolute error (MAE) of 0.8 MPa for the predictions, outperforming the remaining models. XML highlighted the true impact of GO and the remaining variables, emphasising that the presence of GO significantly improves compressive strength. The optimum GO content and optimum superplasticiser content observed from the experimental work agreed with results obtained from the XML analysis, showing the consistency of the explanation with the experimental results. This implies the superiority of using XML methods in concrete mix design applications in industry.

Early Prediction of Cardiac Arrest in the Intensive Care Unit Using Explainable Machine Learning: Retrospective Study.

Cardiac arrest (CA) is one of the leading causes of death among patients in the intensive care unit (ICU). Although many CA prediction models with high sensitivity have been developed to anticipate CA, their practical application has been challenging due to a lack of generalization and validation. Additionally, the heterogeneity among patients in different ICU subtypes has not been adequately addressed. This study aims to propose a clinically interpretable ensemble approach for the timely and accurate prediction of CA within 24 hours, regardless of patient heterogeneity, including variations across different populations and ICU subtypes. Additionally, we conducted patient-independent evaluations to emphasize the model's generalization performance and analyzed interpretable results that can be readily adopted by clinicians in real-time. Patients were retrospectively analyzed using data from the Medical Information Mart for Intensive Care-IV (MIMIC-IV) and the eICU-Collaborative Research Database (eICU-CRD). To address the problem of underperformance, we constructed our framework using feature sets based on vital signs, multiresolution statistical analysis, and the Gini index, with a 12-hour window to capture the unique characteristics of CA. We extracted 3 types of features from each database to compare the performance of CA prediction between high-risk patient groups from MIMIC-IV and patients without CA from eICU-CRD. After feature extraction, we developed a tabular network (TabNet) model using feature screening with cost-sensitive learning. To assess real-time CA prediction performance, we used 10-fold leave-one-patient-out cross-validation and a cross-data set method. We evaluated MIMIC-IV and eICU-CRD across different cohort populations and subtypes of ICU within each database. Finally, external validation using the eICU-CRD and MIMIC-IV databases was conducted to assess the model's generalization ability. The decision mask of the proposed method was used to capture the interpretability of the model. The proposed method outperformed conventional approaches across different cohort populations in both MIMIC-IV and eICU-CRD. Additionally, it achieved higher accuracy than baseline models for various ICU subtypes within both databases. The interpretable prediction results can enhance clinicians' understanding of CA prediction by serving as a statistical comparison between non-CA and CA groups. Next, we tested the eICU-CRD and MIMIC-IV data sets using models trained on MIMIC-IV and eICU-CRD, respectively, to evaluate generalization ability. The results demonstrated superior performance compared with baseline models. Our novel framework for learning unique features provides stable predictive power across different ICU environments. Most of the interpretable global information reveals statistical differences between CA and non-CA groups, demonstrating its utility as an indicator for clinical decisions. Consequently, the proposed CA prediction system is a clinically validated algorithm that enables clinicians to intervene early based on CA prediction information and can be applied to clinical trials in digital health.

Developing an Early Warning System for Financial Networks: An Explainable Machine Learning Approach.

Identifying the influential variables that provide early warning of financial network instability is challenging, in part due to the complexity of the system, uncertainty of a failure, and nonlinear, time-varying relationships between network participants. In this study, we introduce a novel methodology to select variables that, from a data-driven and statistical modeling perspective, represent these relationships and may indicate that the financial network is trending toward instability. We introduce a novel variable selection methodology that leverages Shapley values and modified Borda counts, in combination with statistical and machine learning methods, to create an explainable linear model to predict relationship value weights between network participants. We validate this new approach with data collected from the March 2023 Silicon Valley Bank Failure. The models produced using this novel method successfully identified the instability trend using only 14 input variables out of a possible 3160. The use of parsimonious linear models developed by this method has the potential to identify key financial stability indicators while also increasing the transparency of this complex system.

Explainable machine learning for hardness prediction of laser powder bed fused Ti-6Al-4V and assisting in understanding effects of process parameters

Rapid fabrication of the Ti-6Al-4 V alloy with the target hardness via laser powder bed fusion (LPBF) is essential to meet the requirements of specific applications. Herein, the explainable machine learning (xML) models were employed to predict the hardness of LPBF-ed parts and to provide insights into the mechanisms linking process-microstructure-hardness. Results indicate that gradient boosting decision tree (GBDT) model outperforms others on test set, achieving an R square (R2) of 0.963, a mean absolute error (MAE) of 2.059 HV, and a root mean squared error (RMSE) of 3.844 HV. The contributions of process parameters were evaluated using SHapley Additive exPlanations (SHAP) values. The results demonstrate that the scanning speed (v) has the most significant influence on hardness, with the mean |SHAP value| accounting for approximately 51.2 % of the overall value, and shows an overall positive correlation with hardness. Experimental validation is consistent with the prediction results that S6 fabricated under the high v displayed a high hardness of 404.58 HV. The high v results in a large number of dislocations in as-built S6, impeding dislocation slip during loading and thus increasing resistance to deformation. The {10—11}<—1012> compression twins, nano-stacking faults and numerous fine α’-colonies provide extra grain boundary strengthening. Furthermore, the substantial volume fractions of interfaces between randomly oriented α’ martensite act as a strong barrier to dislocation slip transmission, thereby offering effective strengthening. This work offers guidance for predicting properties and exploring the underlying mechanisms of process parameters on material properties.

Which fields drive the h-index?

The measurement of the quality of academic research is often done by means of the h-index measure. Although widely accepted, the h-index has some issues and, in particular, it may depend on the scientific field in which a researcher operates. To date there is not a definitive answer as to whether this difference holds, and to what extent it varies. To fill the gap, we propose to operationaly measure the difference in h-index across the sectors of a relatively homogeneous population of all scientists of a nation. To answer the heterogeneity issue we apply three different explainable machine learning models: linear regression, Poisson regression and tree models. Our results show that the latter two models better explain the data. They show that the only sectors for which a difference in h-index is significant are Physics, Biology and Clinical Sciences.

Unraveling the complex interactions between ozone pollution and agricultural productivity in China's main winter wheat region using an interpretable machine learning framework

Surface ozone has become a significant atmospheric pollutant in China, exerting a profound impact on crop production and posing a serious threat to food security. Previous studies have extensively explored the physiological mechanisms of ozone damage to plants. However, the effects of ozone interactions with other environmental factors, such as climate change, on agricultural productivity at the regional scale, particularly under natural conditions, remain insufficiently understood. In this study, we employed an interpretable machine learning framework, specifically the eXtreme Gradient Boosting (XGBoost) algorithm enhanced by SHapley Additive exPlanations (SHAP), to investigate the influence of ozone and its interactions with environmental factors on crop production in China's primary winter wheat region. Additionally, a structural equation model was developed to elucidate the mechanisms driving these interactions. Our findings demonstrate that ozone pollution exerts a significant negative effect on winter wheat productivity (r = −0.47, P < 0.001), with productivity losses escalating from −12.28 % to −22.09 % as ozone levels increase. Notably, the impact of ozone is spatially heterogeneous, with western Shandong province identified as a hotspot for ozone-induced damage. Furthermore, our results confirm the complexity of the relationship between ozone pollution and agricultural productivity, which is influenced by multiple interacting environmental factors. Specifically, we found that severe ozone pollution, when combined with high aerosol concentrations or elevated temperatures, significantly exacerbates crop productivity losses, although drought conditions can partially mitigate these adverse effects. Our study highlights the importance of incorporating the interactive effects of air pollution and climate change into future crop models. The comprehensive framework developed in this study, which integrates statistical modeling with explainable machine learning, provides a valuable methodological reference for quantitatively assessing the impact of air pollution on crop productivity at a regional scale.

Bio-inspired MXene membranes for enhanced separation and anti-fouling in oil-in-water emulsions: SHAP explainability ML

Optimizing membrane performance for efficient water treatment is crucial for sustainable development and environmental protection, aligning with UN SDGs. This study involves experimental design, statistical reliability of small data, and explainable machine learning (ML) using SHAP (Shapley Additive Explanations). The research uses ML models and statistical tests to ensure data reliability and stationarity and investigate various membranes’ fouling and separation efficiency (MX-CM, PDMX-CM, and SPDMX-CM). Stationarity tests, including the Augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests, revealed that MX-CM is stationary at level (I(0)), while PDMX-CM and SPDMX-CM required first differencing (I(1)) to achieve stationarity. SHAP analysis showed that in the fouling study, higher values of PDMX-CM and MX-CM positively influenced model predictions, with SHAP values of +0.09 for Cycle, −0.06 for PDMX-CM, and −0.06 for MX-CM. In the separation efficiency study, Cycle had a neutral impact (0.00), PDMX-CM had a slight positive effect, and MX-CM had a slight negative impact. These findings highlight the importance of ensuring data stationarity and utilizing SHAP for model explainability in predicting membrane performance. Accurate preprocessing and model interpretation enhance decision-making and optimization in membrane fouling and separation efficiency studies, ensuring robust and reliable ML models.

Sample size effects on landslide susceptibility models: A comparative study of heuristic, statistical, machine learning, deep learning and ensemble learning models with SHAP analysis

In landslide susceptibility assessment (LSA), inventory incompleteness impacts the accuracy of different models to varying degrees. However, this area remains under-researched. This study investigated six LSA models from heuristic, statistical, machine learning and ensemble learning models (analytical hierarchy process (AHP), frequency ratio (FR), logistic regression (LR), Keras based deep learning (KBDL), XGBoost, and LightGBM) across six different sample sizes (100%, 90%, 75%, 50%, 25%, and 10%). Results revealed that XGBoost and LightGBM consistently outperformed other models across all sample sizes. The LR and KBDL models followed, while FR model was the most affected by sample size variations. AHP, an empirical model, remained unaffected by sample size. Through SHapley Additive exPlanations (SHAP) analysis, elevation, NDVI, slope, land use, and distance to roads and rivers emerged as pivotal indicators for landslide occurrences in the study area, suggesting that human activities significantly influence these events. Five time-varying indicators regarding human activity and climate validated this inference, which provides a new method to identify landslide triggering factors, especially in areas of intense human activity. Based on the findings, a comprehensive framework for LSA is proposed to assist landslide managers in making informed decisions. Future research should focus on expanding model diversity to address the effects of sample size, enhancing the adaptability of the LSA framework, deepening the analysis of human activity impacts on landslides using explainable machine learning techniques, addressing temporal inventory incompleteness in LSA, and critically evaluating model sensitivity to sample size variations across multiple disciplines.

Explainable Machine Learning Model to Accurately Predict Protein-Binding Peptides

Enzymes play key roles in the biological functions of living organisms, which serve as catalysts to and regulate biochemical reaction pathways. Recent studies suggest that peptides are promising molecules for modulating enzyme function due to their advantages in large chemical diversity and well-established methods for library synthesis. Experimental approaches to identify protein-binding peptides are time-consuming and costly. Hence, there is a demand to develop a fast and accurate computational approach to tackle this problem. Another challenge in developing a computational approach is the lack of a large and reliable dataset. In this study, we develop a new machine learning approach called PepBind-SVM to predict protein-binding peptides. To build this model, we extract different sequential and physicochemical features from peptides and use a Support Vector Machine (SVM) as the classification technique. We train this model on the dataset that we also introduce in this study. PepBind-SVM achieves 92.1% prediction accuracy, outperforming other classifiers at predicting protein-binding peptides.

Explainable machine learning for financial risk management: two practical use cases

We explore the potential of machine learning (ML) models applied in two financial risk management areas, i.e., credit risk management and financial risk hedging, through two practical use cases. This comparative study starts with the issue of explainability in complex ML models used in peer-to-peer lending for credit risk management. The first use case examines the limitations of using Kernel-SHAP with dependent features and evaluates different methods for estimating these dependencies using the Lending Club dataset. Our results suggest that accounting for feature dependence improves the understanding and robustness of prediction explanations. The second use case investigates a dynamic method for hedging foreign exchange risk in international equity portfolios, emphasizing the importance of accurate forecasts of currency returns. The analysis demonstrates that predictions yielded by ML models can significantly enhance the hedging of portfolios against currency risk. These findings highlight the transformative potential of advanced ML models in financial risk management, showcasing their capability to improve financial risk measurement and management. Further, our study outlines future research directions to advance this field.

Submicron Organic Aerosol Types in the Summertime Arctic: Mixing State, Geographic Distribution, and Drivers

AbstractDuring the 2017 summertime Arctic cruise observation campaigns, we measured over 290,000 individual submicron particles and clustered them into two inorganic classes (dominated by sea salt, accounting for 38.6% by number fraction) and five organic classes (dominated by natural and anthropogenic organics, 61.4%), presenting a distinct difference in geographic distribution. In the high Arctic and marginal ice zone (81.1–84.6°N) compared with the low Arctic (Chukchi Sea, Svalbard, and Iceland, &lt;80°N), ocean‐derived organic aerosols were more prevalent (73.6% vs. 37.1%). Specifically, we found sharp contrasts in the geographic distributions of OC‐Ca (organics internally mixed with calcium, 29.0% vs. 9.4%) and OC‐S (organics internally mixed with sulfate, 3.2% vs. 21.4%). Utilizing an explainable machine learning technique, we inferred that OC‐Ca was driven by wind‐blown sea ice and/or sea ice floes and/or bubble bursting within sea ice leads under low wind speed conditions in the high Arctic, while OC‐S tended to associate with elemental carbon, sulfate, and higher temperatures, potentially originating from combustion emissions at low latitude regions.

Explaining machine learning practice: findings from an engaged science and technology studies project

ABSTRACT The widespread use of machine learning (ML) models for decision-making raises critical concerns about transparency and accountability – to which an increasingly popular solution is ‘Explainable AI’ (XAI). Here, the object of explanation is technically complex models which are difficult or even impossible to explain. In contrast, this paper makes a call to de-centre models as the object of explanation and look towards the network of ‘machine learning practice’ that bring models into being, and use. We explore this term through an ethnographic study, conducted in collaboration with a large financial services company. Drawing on recent STS research, we ask: what would an explanation look like from a position that recognises the emergent and relational nature of machine learning practice, and how might this contribute to greater accountability and responsibility for ML in use? Inspired by the engaged programme in STS, we explore if and how approaching explanation through ML practice can be mobilised to intervene in how explanations are done in organisations. Our empirical analysis shows an ‘ecology’ of multiple, situated and intra-acting explanations for machine learning practice across a range of human and non-human actors in the company. We argue that while XAI is inevitably partial and limited, its value lies in establishing explanations as an imperative in contexts where ML is implicated in decision-making. Overall, our research suggests a need to widen and deepen the search for explanations and explore the opportunities for provisional, relational and collective interrogations over what can (and can’t) be explained about ML practice.

Accelerated discovery of eutectic compositionally complex alloys by generative machine learning

Eutectic alloys have garnered significant attention due to their promising mechanical and physical properties, as well as their technological relevance. However, the discovery of eutectic compositionally complex alloys (ECCAs) (e.g. high entropy eutectic alloys) remains a formidable challenge in the vast and intricate compositional space, primarily due to the absence of readily available phase diagrams. To address this issue, we have developed an explainable machine learning (ML) framework that integrates conditional variational autoencoder (CVAE) and artificial neutral network (ANN) models, enabling direct generation of ECCAs. To overcome the prevalent problem of data imbalance encountered in data-driven ECCA design, we have incorporated thermodynamics-derived data descriptors and employed K-means clustering methods for effective data pre-processing. Leveraging our ML framework, we have successfully discovered dual- or even tri-phased ECCAs, spanning from quaternary to senary alloy systems, which have not been previously reported in the literature. These findings hold great promise and indicate that our ML framework can play a pivotal role in accelerating the discovery of technologically significant ECCAs.