Data Generation Strategy Research Articles

Rapid lithological mapping using multi-source remote sensing data fusion and automatic sample generation strategy

ABSTRACT The advancement of remote sensing technology aids geologists in obtaining lithological maps more quickly, comprehensively, and accurately. However, key challenges in lithological mapping include the limited spectral information from individual sensors and the difficulties in visually interpreting lithological samples. In this study, we integrated 241 scenes of optical data and 106 scenes of radar data on the Google Earth Engine (GEE) platform, proposing a rapid lithological identification framework that combines an automatic lithological sample data generation strategy with multi-source data. Using various machine learning algorithms, we evaluated the classification capabilities of heterogeneous predictive factors, feature optimization algorithms, and object-based algorithms. Results indicate that: (1) Combining optical and radar data improves prediction accuracy, with terrain data further enhancing mapping capabilities; (2) Terrain factors contribute most to classification, but SWIR and TIR bands of optical data are critical for lithological identification; (3) The feature optimization algorithm reduces feature redundancy and efficiency issues from multi-source data, achieving 96.51% accuracy with the optimal feature model, an improvement of 0.1%−2.02% over original features; (4) Object-based algorithms show significant potential in mapping areas with large rock outcrops. This study offers new insights for medium- to large-scale lithological maps and provides essential data support for geological work.

Deep-Learning Empowered Customized Chiral Metasurface for Calibration-Free Biosensing.

As a 2D metamaterial, metasurfaces offer an unprecedented avenue to facilitate light-matter interactions. The current "one-by-one design" method is hindered by time-consuming, repeated testing within a confined space. However, intelligent design strategies for metasurfaces, limited by data-driven properties, have rarely been explored. To address this gap, a data iterative strategy based on deep learning, coupled with a global optimization network is proposed, to achieve the customized design of chiral metasurfaces. This methodology is applied to precisely identify different chiral molecules in a label-free manner. Fundamentally different from the traditional approach of collecting data purely through simulation, the proposed data generation strategy encompasses the entire design space, which is inaccessible by conventional methods. The dataset quality is significantly improved, with a 21-fold increase in the number of chiral structures exhibiting the desired circular dichroism (CD) response (>0.6). The method's efficacy is validated by a monolayer structure that is easily prepared, demonstrating advanced sensing abilities for enantiomer-specific analysis of bio-samples. These results demonstrate the superior capability of data-driven schemes in photonic design and the potential of chiral metasurface-based platforms for calibration-free biosensing applications. The proposed approach will accelerate the development of complex systems for rapid molecular detection, spectroscopic imaging, and other applications.

Exploring How Female Scientists are Uplifted in Research Networks in STEM Fields in Public Universities in Uganda

Participation in research networks is inevitable for the female scientists in the Science Technology Engineering and Mathematics (STEM) fields because they provide a platform for collaborative activities. New research communities of women are created to support each other through mentoring, advancing shared research interests, co-organizing conferences with men, participating in the editorial and the publications. It was on this basis that this study explored how female scientists were uplifted by research networks in STEM fields in public universities in Uganda. This study was anchored on the existential phenomenology by Martin Heidegger in the interpretive world view. Eight female scientists in research networks in STEM fields in public universities in Uganda were interviewed and they were selected from the hard applied and hard pure disciplines as classified in Becher and Biglan’s typology of academic disciplines. Using thematic analysis, two themes such as research collaboration activities and capacity development emerged in this study. This qualitative study set out to explore how female scientists were uplifted in research networks in public universities in Uganda.The public universities included Makerere University, Mbarara University of Science and Technology, Gulu University and Busitema University. Both semi-structured interviews and document check data generation strategies were used in this study. In semi-structured interviews, a sample comprising eight female academics with not less than five years in research networks in public universities were selected, while documents such as the annual reports and strategic plans were used. This study concludes that research networks are uplifting the female scientists in STEM fields in public universities in Uganda because they were internationally exposed, networked, funded, research skills were strengthened, and progressed in career. This study concludes that it was important for female scientists to access physical and human resource in their research networks as this increased their research performance and outputs in STEM fields in public universities in Uganda. This study therefore recommends that, to uplift female scientists in research networks in STEM fields in public universities in Uganda, female scientists in the research networks and senior colleagues should act as ambassadors and mentors to young female scientists in STEM fields

Biomass hydrothermal gasification characteristics study: based on deep learning for data generation and screening strategies

Hydrothermal gasification is an import-ant way to utilize biomass resources, and accurate prediction of the biomass hydrothermal gasification process is of great significance for the formulation and optimization of process parameters and equipment. Data is a key factor in machine learning, but due to the high experimental costs associated with hydrothermal gasification processes, acquiring a large amount of experimental data for machine learning modeling is a major challenge. To address this issue, this study proposes a data generation and screening strategy based on generative adversarial networks (GAN). The data generation and screening strategy primarily rely on GAN networks trained with real data as data generators and random forest models trained with real data as data screeners. High-quality synthetic data is selected through screening criteria to augment the dataset. Four machine learning models are used to model the biomass hydrothermal gasification process based on synthetic data to validate this strategy. The results show that, compared to the original data, modeling with synthetic data leads to a significant increase in the evaluation metrics for predicting H2, CH4, and CO2, especially during the testing phase. This indicates the rationality of the proposed data generation and screening strategy.

Mutual-guided scale-aggregation denoising network for seismic noise attenuation

The background noise contained in seismic records contaminate the effective reflection waves and impact the subsequent processes, such as inversion and migration. The properties of seismic noises, such as non-Gaussianity and non-linearity, will be even more complex in challenging exploration environments. Deep-learning techniques are effective in suppressing complex seismic noises and outperform conventional denoising algorithms. Nonetheless, most deep learning networks are designed to extract the features of input data in single-scale only, which leads to inadequate performance when dealing with complicated seismic data. To enhance the denoising capability for seismic noises of deep learning, a novel mutual-guided scale-aggregation denoising network (MSD-Net) is designed to suppress seismic noises by utilizing the multi-scale features of input data. Specifically, the MSD-Net achieves functions including multi-scale feature extraction, fusion, and guidance through information interaction between different scales. Spatial aggregation attention is used in MSD-Net to enhance relevant features, which improves the separation of effective reflection waves and noises further. Additionally, a model-based training data generation strategy is devised to ensure the efficiency of learning and the denoising capability of MSD-Net. Compared to conventional denoising algorithms and typical deep learning networks, MSD-Net shows powerful result in suppressing complex seismic noises and generalization.

Climate change adaptation in Malta: assessment of skills in the built environment

PurposeThe construction industry is one of the major contributors of greenhouse gas emissions (UNEP, 2022), but it provides numerous valuable opportunities to contribute to the reduction of greenhouse gas emissions (IPPC, 2023). Innovation within the industry and the adoption of sustainable practices are key to achieve this. This is possible with the contribution of highly knowledgeable professionals and skilled workers. However, different countries, including Malta are facing skill gaps and resource shortages. The purpose of this paper is to define the existing skill gaps and mismatches in climate change adaptation in the built environment in Malta. The research focuses on the positionality of the professionals working in the sector as they operate within a legal framework guided through various policies, and through their role and responsibility, they are in contact with the other stakeholders in the sector.Design/methodology/approachThe data generation strategy used in this study involved a pragmatic approach through a mixed method research methodology, namely, interviews, a questionnaire and a focus group discussion. The three tier scaffolding approach in the data generation strategy allowed for a progressive validation of the data generated.FindingsThe research indicates that the professionals in the sector are aware of their roles and responsibilities to address climate change in their practices, and they have the disposition to acknowledge the need of continuous professional training. Moreover, this study points out that the professionals in the built environment sector are concerned about other gaps within the industry and not solely by skill gaps. Their responsibility to address climate change through greener solutions is often hindered by fragmentation in the system including bureaucratic procedures, clients’ demands, lack of enforcement and general education of the need to address and adapt climate change among the general public. The professionals appeal for an overhaul in the “culture” of the construction industry through a comprehensive rather than a piecemeal approach. This research puts in the limelight the need of a comprehensive strategy to address skill gaps and mismatches in accordance with the need for climate change adaptation as the ultimate challenge rather than addressing different issues separately within the sector.Originality/valueThe paper is to define the existing skill gaps and mismatches in climate change adaptation in the built environment in Malta. By identifying these skill gaps and mismatches, it would be possible to devise an action plan to iron-out such hindrances as part of a broader climate action adaptation strategy.

Electrocardiogram identification based on data generative network and non-fiducial data processing

Nowadays, the use of biological signals as a criterion for identity recognition has gained increasing attention from various organizations and companies. Therefore, it has become crucial to have a biometric identity recognition method that is fast and accurate. In this paper, we propose a linear electrocardiogram (ECG) data preprocessing algorithm based on Kalman filters for rapid noise data filtering (wavelet transform filtering algorithm). Additionally, we introduce a generative network model called Data Generation Strategy Network (DRCN) based on generative networks. The DRCN is employed to augment training samples for convolutional classification networks, ultimately improving the classification performance of the model. Through the final experiments, our method successfully reduced the average misidentification rate of ECG-based identity recognition to 2.5%, and achieved an average recognition rate of 98.7% for each category, significantly surpassing previous achievements. In the future, this method is expected to be widely applied in the field of ECG-based identity recognition.

A bearing fault diagnosis method with an improved residual Unet diffusion model under extreme data imbalance

As a vital constituent of rotating machinery, rolling bearings assume a pivotal function in ensuring the stable operation of equipment. Recently, deep learning (DL)-based methods have been able to diagnose bearing faults accurately. However, in practical applications, the severe data imbalance caused by the limited availability of fault data compared to the abundance of healthy data poses challenges to the effective training of DL models, leading to a decrease in diagnostic accuracy. In this paper, a bearing fault diagnosis method with the improved residual Unet diffusion model (IResUnet-DM) based on a data generation strategy is proposed to solve the extreme data imbalance. Initially, a deep feature extraction network named improved residual Unet is built, which effectively enhances the information learning ability from vibration signals of the Unet network by one-dimensional residual block and self-attention block. Furthermore, the IResUnet-DM is constructed, which generates vibration signals under extreme data imbalance based on a probability model. The variational bound on the negative log-likelihood of the distribution of generated data was optimized to make the generated data similar to the real data distribution. Finally, wide deep convolutional neural network and one-dimensional ResNet classification networks were used for fault identification to verify the validity and generalization of the IResUnet-DM. Experiment results at different data imbalance rates on two bearing datasets demonstrate that the proposed method can effectively improve fault diagnosis accuracy under extreme data imbalances and outperform the comparison method.

Support Services for International Graduate Students at Makerere University

Makerere University is focused on being a research-led university, which necessitates it embrace an internationalisation agenda to fit into the global research network. There is a need to boost the number of international graduate students to help Makerere University realise her vision. This qualitative study explores how international graduate students are supported at Makerere University. The following objective guided it: to explore how international graduate students are supported at Makerere University. The study was anchored in the interpretivism world view, subscribing to the transcendental phenomenology of Edmund Husserl to uncover the support given to international graduate students. Eight international graduate students at Makerere University were interviewed and selected from different academic disciplines categorised according to Becher and Biglan's classification of academic disciplines. Both unstructured interviews and documentary check data generation strategies were used in this study. Data analysis followed a thematic approach whereby various forms of support were provided to international graduate students at Makerere University, and how it is provided was identified. It was concluded that those international graduate students received non-academic and academic support at Makerere University. This study recommends that Makerere University Management provide financial, human, and physical resources to the International Office as ways of facilitating international graduate students to complete their study programs and attracting many more to join the University. The International Office may work with the Office of the Dean of Students to organise events for graduate students to facilitate interactions between international and domestic students, and they may also put in place peer mentors for new international graduate students to foster a sense of community. This study recommends that the International Office work very closely with the schools in the different Colleges of Makerere University, especially establishing a desk officer in each of the Dean's Office to handle issues of international students as this may improve their numbers at Makerere University. Finally, the university management should strengthen and streamline the activities of the International Office at Makerere University

An efficient urban flood mapping framework towards disaster response driven by weakly supervised semantic segmentation with decoupled training samples

Despite the proven effectiveness of data-driven deep learning techniques in urban flood mapping, the availability of annotation data remains a critical factor impeding their timeliness in real applications. Recent progress in weakly supervised semantic segmentation (WSSS) presents promising solutions for addressing this limitation. To accomplish prompt and accurate flood mapping in complex urban areas from high-resolution remote sensing images to support disaster management and response, this study makes contributions in three key aspects: weak training data generation, the improvement of WSSS algorithm, and the construction of benchmark datasets. Firstly, we present a novel yet efficient weak training data generation strategy by decoupling the acquisition of positive and negative samples. This strategy enables the rapid generation of block-level weak annotations assisted by pre-flood river data or the segment anything model (SAM) for zero-shot segmentation, thereby alleviating the burden of weak data labeling. Secondly, to enhance the flood detection results in complex urban landscapes based on low-cost weak labels, we design an end-to-end WSSS framework incorporating tree filtering-based structure constraints and a perturbed dual-branch cross self-distillation mechanism. Lastly, to evaluate the performance of the proposed approach, we constructed two large aerial imagery datasets, namely Calgary-Flood and Huston-Flood. These datasets encompass diverse urban land covers and include challenging scenarios with extensive shadows, providing a robust benchmark for evaluating our method against various urban environments. Experimental results demonstrate that our weak data annotation strategy substantially enhances efficiency. Additionally, the proposed WSSS framework exhibits superior performance in comparison to existing state-of-the-art methods, particularly in terms of edge delineation and algorithmic stability. The advancements in weak data annotation strategy, end-to-end model architecture, and benchmark dataset development in this study collectively exploit the potential of the weakly supervised paradigm for rapid flood mapping in urgent situations. The datasets and code will be publicly available at (https://github.com/YJ-He/Flood_Mapping_WSSS).

Influences of non-landslide sample selection strategies on landslide susceptibility mapping by machine learning

Landslide susceptibility mapping is crucial in mitigating the risk of regional landslide hazards. The main objectives of this study are to design and analyze the impact of three non-landslide data generation methods on landslide susceptibility mapping. The specific non-landslide data selection methods are the random generation of non-landslide points through different distance landslide buffers, by selecting areas with low landslide influence in the landslide condition factor layer, and based on the results of the information value model partitioning by selecting low to medium susceptibility zones. There were 14 landslide condition factors used for the landslide susceptibility mapping, and correlations between landslides and condition factors were analyzed using the Gini index. The 70% dataset was modelled using a logistic regression model as well as an artificial neural network model. Finally, statistical metrics and AUC values were used for the 30% validation data for comparing the predictive performances of different non-landslide data generation strategies in the model. Based on the validation results, models A6 and B6 performed the best, with respective AUC values of 0.997 and 0.995. The findings show that the non-landslide samples generated using the low susceptibility interval of the information value model have the best performance.

Synthetic data generation method for data-free knowledge distillation in regression neural networks

Knowledge distillation is the technique of compressing a larger neural network, known as the teacher, into a smaller neural network, known as the student, while still trying to maintain the performance of the larger neural network as much as possible. Existing methods of knowledge distillation are mostly applicable for classification tasks. Many of them also require access to the data used to train the teacher model. To address the problem of knowledge distillation for regression tasks under the absence of original training data, previous work has proposed a data-free knowledge distillation method where synthetic data are generated using a generator model trained adversarially against the student model. These synthetic data and their labels predicted by the teacher model are then used to train the student model. In this study, we investigate the behavior of various synthetic data generation methods and propose a new synthetic data generation strategy that directly optimizes for a large but bounded difference between the student and teacher model. Our results on benchmark and case study experiments demonstrate that the proposed strategy allows the student model to learn better and emulate the performance of the teacher model more closely.

Genetic variation in Loudetia simplex supports the presence of ancient grasslands in Madagascar

Societal Impact StatementRecognizing Loudetia‐dominated grasslands were widespread prior to human colonization highlights that open ecosystems were and continue to be an important component of Madagascar's biodiversity. A better understanding of the plant species that form grassland ecosystems is necessary for effective land management strategies that support livelihoods, but substantial financial and logistical barriers exist to implementing conservation genetic studies using contemporary genomic tools. Some challenges for population genetic analyses of non‐model polyploids lacking reference genomes can be ameliorated by developing computational resources that leverage a cost‐effective data generation strategy that requires no prior genetic knowledge of the target species. This may benefit conservation programs with small operating budgets while reducing uncertainty compared to status quo microsatellite assays.Summary The extent of Madagascar's grasslands prior to human colonization is unresolved. We used population genetic analyses of a broadly dominant C4 fire‐adapted grass, Loudetia simplex, as a proxy for estimating grassland change through time. We carefully examined the utility of target‐enrichment data for population genetics to make recommendations for conservation genetics. We explored the potential of estimating individual ploidy levels from target‐enrichment data and how assumptions about ploidy could affect analyses. We developed a novel bioinformatic pipeline to estimate ploidy and genotypes from target‐enrichment data. We estimated standard population genetic summary statistics in addition to species trees and population structure. Extended Bayesian skyline plots provided estimates of population size through time for empirical and simulated data. All Malagasy L. simplex individuals sampled in this study formed a clade and possibly indicated an ancestral Central Highland distribution of 800 m in altitude and above. Demographic models suggested grassland expansions occurred prior to the Last Interglacial Period and supported extensive grasslands prior to human colonization. Though there are limitations to target‐enrichment data for population genetic studies, we find that analyses of population structure are reliable. Genetic variation in L. simplex supports widespread grasslands in Madagascar prior to the more recent periods of notable paleoclimatic change. However, the methods explored here could not differentiate between paleoclimatic change near the Last Glacial Maximum and anthropogenic effects. Target‐enrichment data can be a valuable tool for analyses of population structure in the absence a reference genome.

A two-step deep learning method for 3DCT-2DUS kidney registration during breathing

This work proposed KidneyRegNet, a novel deep registration pipeline for 3D CT and 2D U/S kidney scans of free breathing, which comprises a feature network, and a 3D–2D CNN-based registration network. The feature network has handcrafted texture feature layers to reduce the semantic gap. The registration network is an encoder-decoder structure with loss of feature-image-motion (FIM), which enables hierarchical regression at decoder layers and avoids multiple network concatenation. It was first pretrained with a retrospective dataset cum training data generation strategy and then adapted to specific patient data under unsupervised one-cycle transfer learning in onsite applications. The experiment was performed on 132 U/S sequences, 39 multiple-phase CT and 210 public single-phase CT images, and 25 pairs of CT and U/S sequences. This resulted in a mean contour distance (MCD) of 0.94 mm between kidneys on CT and U/S images and MCD of 1.15 mm on CT and reference CT images. Datasets with small transformations resulted in MCDs of 0.82 and 1.02 mm, respectively. Large transformations resulted in MCDs of 1.10 and 1.28 mm, respectively. This work addressed difficulties in 3DCT-2DUS kidney registration during free breathing via novel network structures and training strategies.

Institutional Support Strategies for Academic Career Progression among Female Lecturers: A Case of Kyambogo University of Uganda

Supporting the academic career progression of female lecturers has a positive impact on gender equality within society because universities play a big role in shaping society and building active citizens based on the fact many people who take up powerful roles in society have a background in higher education. It can also enhance their participation in higher education leadership because leadership qualities and potential in higher education are tied to academic rank. However, there is a low academic career progression of female lecturers at Kyambogo University. This qualitative study set out to examine the support strategies employed by the Kyambogo University of Uganda to advance the academic career progression of female lecturers. The study was positioned within an interpretative paradigm and employed a phenomenological approach to uncover the female academics’ experiences. Both semi-structured interviews and documentary analysis data generation strategies were used in this study. In semi-structured interviews, a sample comprising eight female academics with not less than five years at the rank of a lecturer at Kyambogo University were selected, while documents such as the Human Resources Manual which enshrines the policy that guides the promotion of staff in the university and the Gender Policy were analysed. Data were thematically analysed according to phenomenological analysis such as strategies bracketing, clustering into themes, textual description, structural description, and textual-structural essence of the study. The findings which emerged from the study indicate that though the university has some support measures in place, the majority of female academics at Kyambogo University regard the available support strategies as inadequate for instance, most of them have inadequate skills in academic writing, there is inadequate funding of female academics, gender neutral promotional policies, unsatisfactory work environment and scarcity of workshops. The findings further indicate limited women’s academic networks and inadequate mentorship collaboration among academics as the key factors that hamper female lecturers’ academic career progression. The study recommends, among others, academic writing training, adequate funding, and redesigning promotion policies to suit female academic progression.

Predicting continuum breakdown with deep neural networks

The multi-scale nature of gaseous flows poses tremendous difficulties for theoretical and numerical analysis. The Boltzmann equation, while possessing a wider applicability than hydrodynamic equations, requires significantly more computational resources due to the increased degrees of freedom in the model. The success of a hybrid fluid-kinetic flow solver for the study of multi-scale flows relies on accurate prediction of flow regimes. In this paper, we draw on binary classification in machine learning and propose the first neural network classifier to detect near-equilibrium and non-equilibrium flow regimes based on local flow conditions. Compared with classical semi-empirical criteria of continuum breakdown, the current method provides a data-driven alternative where the parameterized implicit function is trained by solutions of the Boltzmann equation. The ground-truth labels are derived rigorously from the deviation of particle distribution functions and the approximations based on the Chapman-Enskog ansatz. Therefore, no tunable parameter is needed in the criterion. Following the entropy closure of the Boltzmann moment system, a data generation strategy is developed to produce training and test sets. Numerical analysis shows its superiority over simulation-based samplings. A hybrid Boltzmann-Navier-Stokes flow solver is built correspondingly with an adaptive partition of local flow regimes. Numerical experiments including the one-dimensional Riemann problem, shear flow layer, and hypersonic flow around a circular cylinder are presented to validate the current scheme for simulating cross-scale and non-equilibrium flow physics. The quantitative comparison with a semi-empirical criterion and benchmark results demonstrates the capability of the current neural classifier to accurately predict continuum breakdown. The code for the data generator, hybrid solver, and neural network implementation is available in the open source repositories [1,2].

Learning From Noisy Labels for MIMO Detection With One-Bit ADCs

This letter presents a data detection method for multiple-input multiple-output systems with one-bit analog-to-digital converters. The basic idea is to learn the likelihood function of the system from training samples. To this end, a training data generation strategy is first proposed, which labels a one-bit received signal with a symbol index determined by channel-based data detection. This strategy requires no extra training overhead beyond pilot symbols for channel estimation, but leads to noisy labels due to data detection errors. For accurate learning from the noisy labels, an expectation-maximization algorithm is also developed. This algorithm learns both the likelihood function and the transition probability from each noisy label to a true label. Numerical results demonstrate that the presented method performs similar to the optimal maximum likelihood detection.

Joint Coherent and Non-Coherent Detection and Decoding Techniques for Heterogeneous Networks

Cellular networks that are traditionally designed for human-type communication (HTC) have the potential to provide cost effective connectivity to machine-type communication (MTC). However, MTC is characterized by unprecedented traffic in cellular networks, thus posing a challenge to its successful incorporation. In this work, we propose a unified framework for amicable coexistence of MTC and HTC. We consider a heterogeneous network where machine-type devices coexist with enhanced mobile broadband (eMBB) devices and propose transceiver techniques that promote efficient signal recovery from these devices. For this, we present an eMBB pilot and MTC data generation strategy that facilitates joint coherent decoding of eMBB data and non-coherent decoding of MTC data. Furthermore, we assess the feasibility of coexistence using receiver operating characteristics, outage probability, and normalized mean square error (NMSE). Our numerical results reveal that a harmonious coexistence of the heterogeneous services can be achieved with properly configured average signal-to-noise ratios and pilot length.

A framework for selecting data generation strategies in qualitative health research studies.

Qualitative health research has the potential to answer important applied health research questions to inform nutrition and dietetics practice, education and policy. Qualitative health research is a distinct subdiscipline of qualitative inquiry that purposefully draws upon the context of healthcare and emphasises health and wellness. Qualitative health research is defined by two parameters: (1) the focus of the study and (2) the methods used. When considering the methods to be used, decisions are required about the type of data to be generated (e.g., transcripts, images and notes) and the process involved in data generation (e.g., interviews, elicitation strategies and observations) to answer the research question(s). Drawing upon examples from nutrition and dietetics literature, this paper provides a framework to support decision-making for nutrition and dietetics researchers and clinician researchers designing conducting qualitative health research. The guiding questions of the framework include: What types of data will be generated? Who is involved in data generation? Where will data generation occur? When will data generation occur? How will data be recorded and managed? and How will participants' and researchers' emotional safety be promoted? Questions about the types of data, those involved, where and when, as well as how safety can be maintained in data generation, not only support a more robust design and description of data generation methods but also keep the person at the centre of the research.

KCV-B: BOOTSTRAP WITH CROSS-VALIDATION FOR DEEP LEARNING MODEL DEVELOPMENT, ASSESSMENT AND SELECTION

Abstract. This study investigates the inability of two popular data splitting techniques: train/test split and k-fold cross-validation that are to create training and validation data sets, and to achieve sufficient generality for supervised deep learning (DL) methods. This failure is mainly caused by their limited ability of new data creation. In response, the bootstrap is a computer based statistical resampling method that has been used efficiently for estimating the distribution of a sample estimator and to assess a model without having knowledge about the population. This paper couples cross-validation and bootstrap to have their respective advantages in view of data generation strategy and to achieve better generalization of a DL model. This paper contributes by: (i) developing an algorithm for better selection of training and validation data sets, (ii) exploring the potential of bootstrap for drawing statistical inference on the necessary performance metrics (e.g., mean square error), and (iii) introducing a method that can assess and improve the efficiency of a DL model. The proposed method is applied for semantic segmentation and is demonstrated via a DL based classification algorithm, PointNet, through aerial laser scanning point cloud data.