Combining Public Machine Learning Models by Using Word Embedding for Human Activity Recognition

We present a novel Cascade Reliability Framework (CRF) that integrates two independent cascade layers of reliability (i.e., variational temperature scaling and conformal prediction) with a pre-trained Machine Learning (ML) model in order to provide clinicians with a more reliable and tunable tool for early-stage diagnosis of Colorectal Cancer (CRC) polyps. The conformal prediction layer generates predictive sets that are guaranteed to contain the true polyp type with an adjustable error rate tuned by clinicians, while the confidence calibration generates meaningful confidence estimates for each predicted label. These two layers provide additional information and an error-tuning-ability for clinicians to assist them in making informed and intuitive decisions considering the outputs of the pre-trained ML model. Utilizing a novel vision-based tactile sensor and unique 3D-printed CRC polyp phantoms, we evaluated the trustworthiness of the proposed architecture and particularly dual outputs of four different types of CRF models, integrated with two different pre-trained ML models (i.e., ResNet18 and Dilated Residual Network) to highlight the model-agnostic feature of the architecture. To thoroughly assess the performance of the proposed approach, we used reliability diagrams and metrics such as accuracy, coverage, and average set size, while also addressing inter-class performance. Results demonstrate that the calibrated CRF models are well capable of handling non-ideal inputs with noise and blur. Moreover, using the conformal prediction with a user-defined error rate and various experiments, we show how clinicians can intuitively interact with a pre-trained ML model to make informed decisions and minimize the risk of CRC polyps misdiagnoses.

Understanding the spatial distribution of soil organic carbon (SOC) content over different climatic regions will enhance our knowledge of carbon gains and losses due to climatic change. However, little is known about the SOC content in the contrasting arid and sub-humid regions of Iran, whose complex SOC–landscape relationships pose a challenge to spatial analysis. Machine learning (ML) models with a digital soil mapping framework can solve such complex relationships. Current research focusses on ensemble ML models to increase the accuracy of prediction. The usual ensemble method is boosting or weighted averaging. This study proposes a novel ensemble technique: the stacking of multiple ML models through a meta-learning model. In addition, we tested the ensemble through rescanning the covariate space to maximize the prediction accuracy. We first applied six state-of-the-art ML models (i.e., Cubist, random forests (RF), extreme gradient boosting (XGBoost), classical artificial neural network models (ANN), neural network ensemble based on model averaging (AvNNet), and deep learning neural networks (DNN)) to predict and map the spatial distribution of SOC content at six soil depth intervals for both regions. In addition, the stacking of multiple ML models through a meta-learning model with/without rescanning the covariate space were tested and applied to maximize the prediction accuracy. Out of six ML models, the DNN resulted in the best modeling accuracies, followed by RF, XGBoost, AvNNet, ANN, and Cubist. Importantly, the stacking of models indicated a significant improvement in the prediction of SOC content, especially when combined with rescanning the covariate space. For instance, the RMSE values for SOC content prediction of the upper 0–5 cm of the soil profiles of the arid site and the sub-humid site by the proposed stacking approaches were 17% and 9% respectively, less than that obtained by the DNN models—the best individual model. This indicates that rescanning the original covariate space by a meta-learning model can extract more information and improve the SOC content prediction accuracy. Overall, our results suggest that the stacking of diverse sets of models could be used to more accurately estimate the spatial distribution of SOC content in different climatic regions.

Membership inference attacks (MIAs) are important measures to evaluate potential risks of privacy leakage from machine learning (ML) models. State-of-the-art MIA defenses have achieved favorable privacy-utility trade-offs using knowledge distillation on split training datasets. However, such defenses increase computational costs as a large number of the ML models must be trained on the split datasets. In this study, we proposed a new MIA defense, called SEDMA, based on self-distillation using model aggregation to mitigate the MIAs, inspired by the model parameter averaging as used in federated learning. The key idea of SEDMA is to split the training dataset into several parts and aggregate multiple ML models trained on each split for self-distillation. The intuitive explanation of SEDMA is that model aggregation prevents model over-fitting by smoothing information related to the training data among the multiple ML models and preserving the model utility, such as in federated learning. Through our experiments on major benchmark datasets (Purchase100, Texas100, and CIFAR100), we show that SEDMA outperforms state-of-the-art MIA defenses in terms of membership privacy (MIA accuracy), model accuracy, and computational costs. Specifically, SEDMA incurs at most approximately 3 - 5% model accuracy drop, while achieving the lowest MIA accuracy in state-of-the-art empirical MIA defenses. For computational costs, SEDMA takes significantly less processing time than a defense with the state-of-the-art privacy-utility trade-offs in previous defenses. SEDMA achieves both favorable privacy-utility trade-offs and low computational costs.

12074 Background: Skeletal-related events (SREs) are detrimental clinical events in bone metastasis from advanced cancer, defined by pathologic fracture, spinal cord compression, and inevitable surgical or radiational intervention to the bone. Given their negative impact on quality of life and prognosis and interpatient heterogeneity in the SRE risk, accurate identification of patients with high SRE risk is critical. Methods: The patient-level data from three randomized clinical trials that administered zoledronic acid (ZA) to patients with bone-metastatic breast cancer, castration-resistant prostate cancer (CRPC), and other types of cancer were analyzed. (N = 460, 452, and 315, respectively) Machine learning (ML) models to predict SREs within 18 months (breast cancer), 12 months (CRPC), and 9 months (other cancers) were developed based on more than 40 baseline clinical and laboratory data. Seven ML algorithms and five feature selection methods were utilized to develop multiple models. The ML model with the best performance was identified based on the F1 score and the area under the receiver operating characteristic curve (AUC-ROC) for each cancer type and interrogated for important features with Shapley additive explanations. Lastly, the ML models’ ability to stratify patients by the cumulative SRE risk was evaluated by calculating hazard ratio (HR) with Cox-proportional hazards models. Results: Among the multiple ML models developed with different algorithms and feature selection methods, the model developed utilizing the random forests algorithm and the Boruta method for selecting features demonstrated the best performance in all types of cancer (F1 0.70, 0.67, and 0.67, and AUC-ROC 0.72, 0.68, and 0.73 for breast cancer, CRPC, and other cancers, respectively). In the ML model for breast cancer, performance status (PS), history of SRE, serum alkaline phosphatase (ALP), history of anti-neoplastic surgery, radiation therapy, and pathologic fracture were included as important features. Serum ALP, albumin, sodium, Gleason scores, and geographic regions were shown to be relevant in the CRPC model. For other cancers, serum ALP, albumin, total protein, phosphorus, red blood cell count, white blood cell count, visceral metastases, and history of arthritis were incorporated in the ML model. The ML model prediction successfully stratified the patients for cumulative SRE risk in all three cohorts. (HR with 95% confidence interval: 2.43 [1.86 – 3.18], 1.92 [1.51 – 2.45], and 3.06 [2.29 – 4.09] for breast cancer, CRPC, and other types of cancer, respectively). Conclusions: ML models incorporating baseline clinical and laboratory data can identify patients with bone metastasis on ZA harboring a high SRE risk.

As the number of pre-trained machine learning (ML) models is growing exponentially, data reduction tools are not catching up. Existing data reduction techniques are not specifically designed for pre-trained model (PTM) dataset files. This is largely due to a lack of understanding of the patterns and characteristics of these datasets, especially those relevant to data reduction and compressibility. This paper presents the first, exhaustive analysis to date of PTM datasets on storage compressibility. Our analysis spans different types of data reduction and compression techniques, from hash-based data deduplication, data similarity detection, to dictionary-coding compression. Our analysis explores these techniques at three data granularity levels, from model layers, model chunks, to model parameters. We draw new observations that indicate that modern data reduction tools are not effective when handling PTM datasets. There is a pressing need for new compression methods that take into account PTMs' data characteristics for effective storage reduction. Motivated by our findings, we design Elf, a simple yet effective, error-bounded, lossy floating-point compression method. Elf transforms floating-point parameters in such a way that the common exponent field of the transformed parameters can be completely eliminated to save storage space. We develop Elves, a compression framework that integrates Elf along with several other data reduction methods. Elves uses the most effective method to compress PTMs that exhibit different patterns. Evaluation shows that Elves achieves an overall compression ratio of 1.52×, which is 1.31×, 1.32× and 1.29× higher than a general-purpose compressor (zstd), an error-bounded lossy compressor (SZ3), and the uniform model quantization, respectively, with negligible model accuracy loss.

Shale caprocks, often organic-rich, play a significant role in hydrogen sequestration, but their wettability is typically altered from water-wet to H2-wet due to the organic matter, compromising their sealing performance. Silica nanofluids can restore water-wet conditions, which is vital for effective containment of H2 in geological formations. However, laboratory experiments to test the effects of nanomaterials on wettability and interfacial tension (IFT) are resource-intensive and time-consuming. To address this challenge, data-driven machine learning (ML) models offer a promising alternative, allowing for the prediction of key UHS parameters—such as advancing and receding contact angles without the need for extensive laboratory testing. We have applied multiple ML models, including gradient boosting, extreme gradient boosting (XGboost), Catboost, random forest, and extra trees to predict the wettability of organic rich shale samples treated with silica nanofluids. The ML models were trained on a comprehensive dataset generated through laboratory experiments conducted under a wide range of pressure (0.5 – 1600psi) and temperature (298-323 K) conditions, using brine (1 wt.% KCl + 2 wt.% NaCl) and varying silica nanofluid concentrations (0.05–1.0 wt.%). Advanced visualization techniques, including swarm maps, box plots, and heatmap plots, were utilized to analyze the experimental dataset thoroughly. To enhance the generalization capabilities of the ML models, we employed Bayesian optimization for hyperparameter tuning, along with k-fold cross-validation and grid-search optimization techniques. Both statistical and graphical analyses were performed to show the reliability and performance of the models. The results demonstrated that the implemented ML models, particularly the XGboost and Catboost, accurately predicted the wettability behavior under various operating conditions. The training and testing average R² scores were above 0.998 and 0.949, respectively, confirming the high accuracy of the predictions. These findings highlight the reliability of the ML algorithms in predicting the complex interaction between shale, brine, and H2. Further investigation into the feature importance revealed that nanomaterials had the greatest influence on the wettability of the ternary system. The silica nanofluid concentrations played a key role in modifying wettability, with varying impacts on the receding contact angles, emphasizing the effect of nanomaterial treatment. The model's predictions were consistent with experimental results, suggesting that geological conditions, including pressure and temperature, have a profound influence on rock mineral wettability, directly impacting H2 geo-storage capacity. This study is the first to employ data-driven ML models to predict critical UHS parameters in organic-rich shales treated with silica nanofluids. The ability of the ML models to predict UHS parameters across a wide range of conditions provides a valuable tool for optimizing UHS system designs, offering a more efficient and scalable approach compared to traditional experimental methods.

In this work, we present a machine learning (ML) model for predicting the compressive yield strength of multi-principal element alloys (MPEAs). The model uses physical and thermodynamic alloy descriptors that capture various aspects of the alloying process such as lattice distortion, chemical interaction and intermetallic formation, and was used to explore yield strength variations in a variety of pseudo-ternary alloy systems viz. Zr x -(NbTa) y -(MoW) 1− x − y , W x -Ti y -(HfNbTaZr) 1− x − y and (MoNb) x -(AlCr) y -(FeNi) 1− x − y wherein it was validated along continuous composition pathways through comparison with experimentally observed yield strength values. The Compositional-Stimulus and Model-Response framework was used to decode the decision-making process of the ML model for obtaining alloy specific insights. Further, the model predictions were compared with hardness and phase predictions from other pre-trained ML models to establish consistency between different ML models. The model enables high-throughput screening for targeted yield strength and identifies key features for capturing yield strength variations in MPEAs. Graphical abstract

• ML models for predicting SCM concretes compressive strength developed • Multiple datasets devised from a global dataset were used for training • Multiple ML models used in conjunction with multiple datasets for best results • Accurate prediction results obtained for all SCM concrete types studied • Prediction uncertainty bounds established by conformal prediction analysis Supplementary cementitious materials (SCM) used as partial replacements of cement can reduce carbon emissions and construction costs. The amount and types of SCM used directly influence the compressive strength of concrete, which is a crucial indicator of concrete performance. However, predicting compressive strength is challenging for SCM-based concretes due to their complex compositions, especially when multiple types of SCM are used. This study used five machine learning (ML) models, namely artificial neural network (ANN), random forest (RF), extra trees (ET), extreme gradient boosting (XGB), and adaptive boosting (AdaBoost) to predict the 28-day compressive strength of SCM concretes incorporating fly ash (FA), silica fume (SF), and ground granulated blast furnace slag (GGBFS). Firstly, a global dataset comprising 1456 data points was organized into seven individual and six combined datasets for training of the five ML models. After this, the performances of the ML models’ predictions using individual and combined datasets were compared using root mean squared error (RMSE) and coefficient of determination (R 2 ) to identify the optimal ML model-dataset configurations for each type of SCM-based concrete. It is found that in general, XGB models developed using the combined datasets demonstrated superior prediction accuracy, achieving an R² value exceeding 0.8. Both Shapley Additive Explanation (SHAP) and partial dependence plots (PDPs) were employed to illustrate the effects of input variables on compressive strength. Finally, uncertainty quantification using the Jackknife method combined with a normalized nonconformity measure was employed to provide accurate uncertainty bounds for compressive strength predictions.

Social media platforms provide valuable insights into mental health trends by capturing user-generated discussions on conditions such as depression, anxiety, and suicidal ideation. Machine learning (ML) and deep learning (DL) models have been increasingly applied to classify mental health conditions from textual data, but selecting the most effective model involves trade-offs in accuracy, interpretability, and computational efficiency. This study evaluates multiple ML models, including logistic regression, random forest, and LightGBM, alongside DL architectures such as ALBERT and Gated Recurrent Units (GRUs), for both binary and multi-class classification of mental health conditions. Our findings indicate that ML and DL models achieve comparable classification performance on medium-sized datasets, with ML models offering greater interpretability through variable importance scores, while DL models are more robust to complex linguistic patterns. Additionally, ML models require explicit feature engineering, whereas DL models learn hierarchical representations directly from text. Logistic regression provides the advantage of capturing both positive and negative associations between features and mental health conditions, whereas tree-based models prioritize decision-making power through split-based feature selection. This study offers empirical insights into the advantages and limitations of different modeling approaches and provides recommendations for selecting appropriate methods based on dataset size, interpretability needs, and computational constraints.

The majority of ungauged regions around the world are in protected areas and rivers with non-perennial flow regimes, which are vital to water security and conservation. There is a limited amount of ground data available in such regions, making it difficult to obtain streamflow information. This study examines how in situ streamflow datasets in data- rich regions can be used to extrapolate streamflow information into regions with poor data availability. These data-rich regions include North America (987 catchments), South America (813 catchments), and Western Europe (457 catchments). South Africa and Central Asia are defined as data-poor regions. We obtained 81 catchments and 133 catchments for these two data-poor regions, respectively, and assumed they are pseudo ungauged regions for our analysis. We trained machine learning (ML) algorithms using climate and catchments attributes input variables in data-rich (i.e., source) regions and analyzed the possibility of using these pre-trained ML models to estimate climatological monthly streamflow over data-poor (i.e., target) regions. We found that including diverse climate and catchment attributes in training data sets can greatly improve ML algorithms' performance regardless of significant geographical distance between input datasets. The pre-trained ML models over North America and South America could be used effectively to estimate streamflow over data-poor regions. This study provides insight into the selection of input datasets and ML algorithms with different sets of hyperparameters for a geographic streamflow extrapolation.

Designing of potential polymer donors with lower reorganization energy and chemical space exploration in search of materials for efficient organic solar cells

Ensuring cross-device portability of electromagnetic side-channel analysis for digital forensics

Examining the electrical discharge machining (EDM) process is challenging in manufacturing technology due to the complexity of the physical events that take place in the gaps between electrodes. In this study, we examined the EDM process in detail and developed multiple machine learning (ML) models to describe the relationship between the EDM independent (process parameters) and dependent (responses) variables. The collected experimental data was used to train the machine learning models. According to the results, the GPR model outperformed other ML models across different materials, with average RMSE values of 0.9234 and 3.0216 for the material removal rate (MRR) and surface roughness (Sa), respectively, highlighting the effectiveness of ML tools at modeling complex machining processes, such as EDM. In addition, as a practical implication, this study opens the door to employing the developed ML models to predict the EDM process performance.

Deep learning techniques for human activity recognition are gaining popularity due to their effectiveness in identifying intricate tasks and their cost-effectiveness compared to conventional machine learning methods. Human Activity Recognition (HAR) is a research domain concerned with detecting the everyday activities performed by individuals using time-series data captured by sensors. HAR encompasses a wide range of applications, including surveillance, baby monitoring, elderly healthcare, and smart driving. This article provides a brief introduction to the application of deep learning in HAR. It covers the fundamental concepts of CNNs and LSTMs, their strengths in capturing spatial and temporal features, and their integration for enhanced activity recognition. Different approaches are employed in HAR to address problems with efficiency and precision. Conventional human activity recognition (HAR) systems rely on wearable devices like IMUs and stretch sensors to identify different activities. These systems have proven to be effective in recognizing simple user actions like sitting, standing, and walking. However, when it comes to more intricate activities like running, jumping, wrestling, and swinging, sensor-based HAR systems encounter greater misclassification rates due to inaccuracies in sensor readings. These errors significantly impact the accuracy of the HAR system, resulting in suboptimal classification outcomes. In contrast, employing vision-based HAR systems enables improved accuracy in identifying complex activities, leading to enhanced overall performance.

The medical imaging community has embraced Machine Learning (ML) as evidenced by the rapid increase in the number of ML models being developed, but validating and deploying these models in the clinic remains a challenge. The engineering involved in integrating and assessing the efficacy of ML models within the clinical workflow is complex. This paper presents a general-purpose, end-to-end, clinically integrated ML model deployment and validation system implemented at UCSF. Engineering and usability challenges and results from 3 use cases are presented. A generalized validation system based on free, open-source software (OSS) was implemented, connecting clinical imaging modalities, the Picture Archiving and Communication System (PACS), and an ML inference server. ML pipelines were implemented in NVIDIA’s Clara Deploy framework with results and clinician feedback stored in a customized XNAT instance, separate from the clinical record but linked from within PACS. Prospective clinical validation studies of 3 ML models were conducted, with data routed from multiple clinical imaging modalities and PACS. Completed validation studies provided expert clinical feedback on model performance and usability, plus system reliability and performance metrics. Clinical validation of ML models entails assessing model performance, impact on clinical infrastructure, robustness, and usability. Study results must be easily accessible to participating clinicians but remain outside the clinical record. Building a system that generalizes and scales across multiple ML models takes the concerted effort of software engineers, clinicians, data scientists, and system administrators, and benefits from the use of modular OSS. The present work provides a template for institutions looking to translate and clinically validate ML models in the clinic, together with required resources and expected challenges.