More Like this

1. Creating and Analyzing a Multimedia Dataset for Building Energy Efficiency Estimation

   Chen, G Q; Kashyap, N; Zhang, Z; Kucheva, Y; Bobker, M; Zhu, Z (March 2024, International Conference on SMART MULTIMEDIA)

   This paper presents the results of a research that created and analyzed a Multimedia dataset for building energy efficiency estimation. First a new Multimedia Building Energy Efficiency (MMBEE) dataset was created from publicly available data. This work then explored the use of the window-to-wall ratio (WWR) information from building facade images and integrated it with traditional tabular data to create new training data, in order to predict building energy efficiency measures. Finally, we discuss potential applications and future research directions in using the MMBEE dataset for building energy efficiency prediction. Throughout the paper, a number of important processes and analyses were performed, which include feature selection, data correlation analysis, WWR extraction, and comparison of deep network and random forest models in building energy efficiency estimation. From this first attempt at using the Multimedia dataset for building energy efficiency estimation, we found the performances of deep models were better than traditional models such as random forest. We also found that there was an optimal point of what features shall be used for the prediction. Nonetheless, the incorporation of the current WWR estimation results did not yield the anticipated enhancement in estimation performance. Subsequently, a comprehensive investigation was conducted to ascertain potential contributing factors, and several avenues for future research were identified to enhance the predictive utility of the WWR feature. 

   more » « less
2. Real-Time Posture Classification on Resource-Constrained Runtime Environments: A Comparative Study of Traditional and Deep Learning Models

   https://doi.org/10.1109/SoutheastCon56624.2025.10971620  

   Shejuti, Zarin; Fuad, Muztaba (March 2025, IEEE)

   While machine learning models perform well on offline data, assessing their performance in real-world, resource-constrained environments-considering accuracy, prediction time, power consumption, and memory usage-is crucial for practical applications. This research implements a mobile-based Human Activity Recognition solution to classify three postures-sitting, standing, and walking-using smartphone sensors, specifically accelerometer, gyroscope, and magnetometer. Time-domain features extracted from these sensors were used, with Random Forest employed for feature selection. One traditional machine learning model, Logistic Regression, and one deep learning model, Convolutional Neural Network, were trained and deployed via an Android application for real-time evaluation. While the Convolutional Neural Network achieved higher accuracy and better memory efficiency, Logistic Regression demonstrated faster prediction times during real-time use. Both models showed reduced accuracy for standing and walking postures in real-world conditions, emphasizing the challenges of deploying machine learning models in dynamic environments. This study highlights the importance of evaluating machine learning models in real-world settings to ensure reliability and efficiency, particularly in resource-constrained environments. 

   more » « less
3. Network-based drug sensitivity prediction

   https://doi.org/10.1186/s12920-020-00829-3  

   Ahmed, Khandakar Tanvir; Park, Sunho; Jiang, Qibing; Yeu, Yunku; Hwang, TaeHyun; Zhang, Wei (December 2020, BMC Medical Genomics)

   null (Ed.)

   Abstract Background Drug sensitivity prediction and drug responsive biomarker selection on high-throughput genomic data is a critical step in drug discovery. Many computational methods have been developed to serve this purpose including several deep neural network models. However, the modular relations among genomic features have been largely ignored in these methods. To overcome this limitation, the role of the gene co-expression network on drug sensitivity prediction is investigated in this study. Methods In this paper, we first introduce a network-based method to identify representative features for drug response prediction by using the gene co-expression network. Then, two graph-based neural network models are proposed and both models integrate gene network information directly into neural network for outcome prediction. Next, we present a large-scale comparative study among the proposed network-based methods, canonical prediction algorithms (i.e., Elastic Net, Random Forest, Partial Least Squares Regression, and Support Vector Regression), and deep neural network models for drug sensitivity prediction. All the source code and processed datasets in this study are available at https://github.com/compbiolabucf/drug-sensitivity-prediction . Results In the comparison of different feature selection methods and prediction methods on a non-small cell lung cancer (NSCLC) cell line RNA-seq gene expression dataset with 50 different drug treatments, we found that (1) the network-based feature selection method improves the prediction performance compared to Pearson correlation coefficients; (2) Random Forest outperforms all the other canonical prediction algorithms and deep neural network models; (3) the proposed graph-based neural network models show better prediction performance compared to deep neural network model; (4) the prediction performance is drug dependent and it may relate to the drug’s mechanism of action. Conclusions Network-based feature selection method and prediction models improve the performance of the drug response prediction. The relations between the genomic features are more robust and stable compared to the correlation between each individual genomic feature and the drug response in high dimension and low sample size genomic datasets. 

   more » « less
4. Predictive machine learning models trained on experimental datasets for electrochemical nitrogen reduction

   https://doi.org/10.1039/d3dd00151b  

   Rosser, Darik A; Farris, Brianna R; Leonard, Kevin C (April 2024, Digital Discovery)

   Obtaining useful insights from machine learning models trained on experimental datasets collected across different groups to improve the sustainability of chemical processes can be challenging due to the small size and heterogeneity of the dataset. Here we show that shallow learning models such as decision trees and random forest algorithms can be an effective tool for guiding experimental research in the sustainable chemistry field. This study trained four different machine learning algorithms (linear regression, decision tree, random forest, and multilayer perceptron) using different sized datasets containing up to 520 unique reaction conditions for the nitrogen reduction reaction (NRR) on heterogeneous electrocatalysts. Using the catalyst properties and experimental conditions as the features, we determined the ability of each model to regress the ammonia production rate and the faradaic efficiency. We observed that the shallow learning decision tree and random forest models had equal or better predictive power compared to the deep learning multilayer perceptron models and the simple linear regression models. Moreover, decision tree and random forest models enable the extraction of feature importance, which is a powerful tool in guiding experimental research. Analysis of the models showed the complex interaction between the applied potential and catalysts on the effective rate for the NRR. We also suggest some underexplored catalysts–electrolyte combinations to experimental researchers looking to improve both the rate and efficiency of the NRR reaction. 

   more » « less
5. Benchmarking RRAM Crossbar Arrays for Epileptic Seizure Prediction

   https://doi.org/10.1109/MWSCAS60917.2024.10658668  

   Khan, Ahmedul; Varnosfaderani, Shiva Maleki; Alhawari, Mohammad; Tutuncuoglu, Gozde (August 2024, IEEE)

   This paper explores an energy-efficient resistive random access memory (RRAM) crossbar array framework for predicting epileptic seizures using the CHB-MIT electroencephalogram (EEG) dataset. RRAMs have significant potential for in-memory computing, offering a promising solution to overcome the limitations of the traditional Von Neumann architecture. By integrating a domain-specific feature extraction approach and evaluating the optimal RRAM hardware parameters using the NeuroSim+ benchmarking platform, we assess the performance of RRAM crossbars for predicting epileptic seizures. Our proposed workflow achieves accuracy levels above 80% despite the EEG data being quantized to 1-bit, highlighting the robustness and efficiency of our approach for epileptic seizure prediction 

   more » « less