Window Size Research Articles

Clinical Relatedness and Stability of vigiVec Semantic Vector Representations of AdverseEvents and Drugs in Pharmacovigilance.

Individual case reports are essential to identify and assess previously unknown adverse effects of medicines. On these reports, information on adverse events (AEs) and drugs are encoded in hierarchical terminologies. Encoding differences may hinder the retrieval and analysis of clinically related reports relevant to a topic of interest. Recent studies have explored the use of data-driven semantic vector representations to support analysis of pharmacovigilance data. This study aims to evaluate the stability and clinical relatedness of vigiVec, a semantic vector representation for codes of AEs and drugs. vigiVec is a published adaptation to pharmacovigilance of the publicly available Word2Vec model, applied to structured data instead of free text. It provides vector representations for MedDRA® Preferred Terms and WHODrug Global active ingredients, learned from reporting patterns in VigiBase, the WHO global database of adverse event reports for medicines and vaccines. For this study, a 20-dimensional Skip-gram architecture with window size 250 was used. Our evaluation focused on nearest neighbors identified by the cosine similarity of vigiVec vector representations. Clinical relatedness was measured through term intruder detection, whereby a medical doctor was tasked to identify a randomly selected term-the intruder-includedamong the four nearest neighbors to a specific AE or drug. Stability was measured as the average overlap in the ten nearest neighbors for each AE or drug, in repeated fittings of vigiVec. Among the ten nearest neighbors, 1.8 AEs on average belonged to the same MedDRA High Level Term (HLT; e.g., coagulopathies), and 1.3 drugs belonged to the same Anatomical Therapeutic Chemical level 3 (ATC-3; e.g., opioids). In the intruder detection task, when neighbors and intruders were both chosen from the same HLT, the intruder detection rate was 46%. When selected from different HLTs, it was 79%. Byrandom chance, we should expect 20%(1 in 5). Corresponding rates for drugs were 42% in same ATC-3 and 65% in different ATC-3. The stability of nearest neighbors was 80% for AEs and 64% for drugs. Nearest neighbors identified with vigiVec are stable and show high level of clinical relatedness. They are often from different parts of the existing hierarchies and complement these.

Quality control of DEMs using check surfaces

ABSTRACT This paper proposes the use of check surfaces or ‘patches’ for the assessment of Digital Elevation Model (DEM) elevations and derived variables (e.g. slope and aspect). We work with grid-type DEMs, and the patches are implemented through square windows of different sizes. Since the data included in a patch are autocorrelated, a trial is performed under controlled conditions of simulation, and then the same simulation method is applied to a DEM data set. Using the Kolmogorov–Smirnov statistic, the observed distribution functions of the results for different sample sizes (20, 50, 100) and different patch sizes (3 × 3, 5 × 5, 9 × 9, 11 × 11, 15 × 15, 19 × 19) and for a sample of points are compared against the true population distribution function. The trial based on autocorrelated synthetic data created following a geostatistical model (exponential law) outputs a clear result regarding the influence of intra-patch autocorrelation. The lower this autocorrelation is, the better the patch performs in terms of sampling efficiency. The same simulation process applied to synthetic data has also been applied to the case of a DEM product. The results obtained allow us to establish an equivalence between sample size when using control patches and control points. The main result of this study is that it allows us to understand the behaviour of samples based on patches and offers us a guide for determining the size of patch-based samples. The most relevant contribution of this proposal is that it opens the possibility of analysing variables derived from elevation data, such as slope and aspect.

Action Recognition in Basketball with Inertial Measurement Unit-Supported Vest.

In this study, an action recognition system was developed to identify fundamental basketball movements using a single Inertial Measurement Unit (IMU) sensor embedded in a wearable vest. This study aims to enhance basketball training by providing a high-performance, low-cost solution that minimizes discomfort for athletes. Data were collected from 21 collegiate basketball players, and movements such as dribbling, passing, shooting, layup, and standing still were recorded. The collected IMU data underwent preprocessing and feature extraction, followed by the application of machine learning algorithms including KNN, decision tree, Random Forest, AdaBoost, and XGBoost. Among these, the XGBoost algorithm with a window size of 250 and a 75% overlap yielded the highest accuracy of 96.6%. The system demonstrated superior performance compared to other single-sensor systems, achieving an overall classification accuracy of 96.9%. This research contributes to the field by presenting a new dataset of basketball movements, comparing the effectiveness of various feature extraction and machine learning methods, and offering a scalable, efficient, and accurate action recognition system for basketball.

Anatomical investigation of safety determining factors for keyhole drilling trajectories for robotic cochlear implant surgery.

Cochlear implants (CI) are the most successful bioprosthesis in medicine probably due to the tonotopic anatomy of the auditory pathway and of course the brain plasticity. Correct placement of the CI arrays, respecting the inner ear anatomy are therefore important. The ideal trajectory to insert a cochlear implant array is defined by an entrance through the round window membrane and continues as long as possible parallel to the basal turn of the cochlea. Image-guided surgery can directly drill with a robotic arm through the mastoid and execute an exact drilling of this precalculated most ideal trajectory. Here, we aim to identify critical anatomical structures determining the safest keyhole drilling trajectory by comparing easy and difficult CI surgeries performed with the HEARO Procedure: a robotic tool for image-guided surgery. Cone-beam computed tomography images of patients who underwent robot-assisted cochlear implantation surgery (RACIS) were included. Three of 25 cases had to be converted to conventional surgery because of the current safety mitigations based on anatomical distance. Radiological images in DICOM format were transferred to dedicated software (OTOPLAN® Cascination GMHB Bern Switzerland) for analyses. Surgical segmentation and previously planned trajectories were analyzed for these 25 cases by comparing cochlear sizes, facial recess sizes, round window sizes, and trajectory angles. In addition, facial recess angle, and cochlear orientation angles were measured with RadiAnt DICOM Viewer (Medixant, Pozan, Poland). Facial recess size, facial recess angle, and distance between the facial nerve and safe trajectory were smaller in patient who converted from robotic surgery to conventional. A significant positive correlation existed between basal turn angle and in-plane angle (p = 0.001, r = 0.859). In addition, there was a significant negative correlation between the basal turn length and the last electrode insertion depth degree (p = 0.007, r = 0.545). In our robotic surgery cases, we demonstrate that the limiting factors of anatomical relationships are constituted by the dimensions of the facial recess and the cochlear orientation. The findings of this study are considered to be a reference for future studies in achieving collision-free trajectory planning in robotic-assisted cochlear implant surgery.

Classifying Food Items During an Eating Occasion: A Machine Learning Approach with Slope Dynamics for Windowed Kinetic Data.

Wearable devices equipped with a range of sensors have emerged as promising tools for monitoring and improving individuals' health and lifestyle. Contribute to the investigation and development of effective and reliable methods for dietary monitoring based on raw kinetic data generated by wearable devices. This study uses resources from the NOTION study. A total of 20 healthy subjects (9 women and 11 men, aged 20-31 years) were equipped with two commercial smartwatches during four eating occasions under semi-naturalistic conditions. All meals were video-recorded, and acceleration data were extracted and analyzed. Food recognition on these features was performed using random forest (RF) models with 5-fold cross-validation. The performance of the classifiers was expressed in out-of-bag sensitivity and specificity. Acceleration along the x-axis and power show the highest and lowest rates of median variable importance, respectively. Increasing the window size from 1 to 5 s leads to a gain in performance for almost all food items. The RF classifier reaches the highest performance in identifying meatballs (89.4% sensitivity and 81.6% specificity) and the lowest in identifying sandwiches (74.6% sensitivity and 72.5% specificity). Monitoring food items using simple wristband-mounted wearable devices is feasible and accurate for some foods while unsatisfactory for others. Machine learning tools are necessary to deal with the complexity of signals gathered by the devices, and research is ongoing to improve accuracy further and work on large-scale and real-time implementation and testing.

A Novel Methodology for Performance Evaluation in Advanced Quality Control

Current global conditions and challenges in industrial manufacturing, marked by dynamism, competition, and the need for responsible resource management, have increased the demand for sustainable manufacturing practices. The integration of Industry 4.0 and the recent development of Industry 5.0 have added dynamism, which has generated profound implications for quality control and process monitoring, focusing mainly on recognising control patterns within the manufacturing environment. This study introduces a novel methodology for evaluating the performance of pattern classification models used in advanced quality control. Our approach incorporates robust performance metrics, early detection, window size, network hyperparameters, and concurrent patterns within a simulated monitoring environment. Unlike previous research, our evaluation methodology addresses the sensitivity of classification models to various factors, emphasising the critical balance between early detection and minimising false alarms. The findings reveal that window size significantly impacts the model’s sensitivity to pattern changes, highlighting that measuring early detection alone is impractical in real-world applications. Furthermore, optimal hyperparameter selection enhances the model’s practical applicability.

The degree of parallel/serial processing affects stimulus-driven and memory-driven attentional capture: Evidence for the attentional window account.

The issue of whether a salient stimulus in the visual field captures attention in a stimulus-driven manner has been debated for several decades. The attentional window account proposed to resolve this issue by claiming that a salient stimulus captures attention and interferes with target processing only when an attentional window is set wide enough to encompass both the target and the salient distractor. By contrast, when a small attentional window is serially shifted among individual stimuli to find a target, no capture is found. Research findings both support and challenge this attentional window account. However, in these studies, the attentional window size was improperly estimated, necessitating a re-evaluation of the account. Here, using a recently developed visual search paradigm, we investigated whether visual stimuli were processed in a parallel or a serial manner. We found significant attentional capture when multiple stimuli were processed in parallel within a large attentional window. By contrast, when a small window had to be serially shifted, no capture was found. We conclude that the attentional window account can be a useful framework to resolve the widespread debate regarding stimulus-driven attentional capture.

Multiscale Sensitivity Analysis of Landscape Fragmentation in Plantation Forests on the Loess Plateau

ABSTRACTEffectively quantifying plantation forest fragmentation is crucial for vegetation restoration and management on the Loess Plateau. Forest area density (FAD) effectively measures fragmentation, with multiscale sensitivity analysis essential for determining the appropriate window scale threshold. This study uses plantation forest data from different geomorphological types (hilly gully and plateau gully region) of the Loess Plateau. Various window sizes are applied to evaluate the FAD threshold curve and its derivative for determining a stable fragmentation threshold, while a structural equation model is employed to analyze the drivers of forest fragmentation. The results are as follows: (1) FAD variability decreases with increasing window size, leveling off after reaching the threshold. (2) Thresholds vary by forest type and county. (3) Two principal components explain 81.27% of the window stabilization threshold variation, with strong correlations between similar fragmentation types. (4) Fragmentation slowed between 2000 and 2022, with hilly gully region more fragmented than plateau gully region. (5) Stand structure mitigates fragmentation, with topographic and climate change influences showing heterogeneity. “Core” maintain stability, while “Islet” drive fragmentation. This study emphasizes the importance of scale‐specific analysis and adaptive management strategies in mitigating forest fragmentation. This study supports optimizing ecological restoration and targeted conservation strategies, promoting sustainable forest management on the Loess Plateau.

Sliding-Window Dissimilarity Cross-Attention for Near-Real-Time Building Change Detection

A near-real-time change detection network can consistently identify unauthorized construction activities over a wide area, empowering authorities to enforce regulations efficiently. Furthermore, it can promptly assess building damage, enabling expedited rescue efforts. The extensive adoption of deep learning in change detection has prompted a predominant emphasis on enhancing detection performance, primarily through the expansion of the depth and width of networks, overlooking considerations regarding inference time and computational cost. To accurately represent the spatio-temporal semantic correlations between pre-change and post-change images, we create an innovative transformer attention mechanism named Sliding-Window Dissimilarity Cross-Attention (SWDCA), which detects spatio-temporal semantic discrepancies by explicitly modeling the dissimilarity of bi-temporal tokens, departing from the mono-temporal similarity attention typically used in conventional transformers. In order to fulfill the near-real-time requirement, SWDCA employs a sliding-window scheme to limit the range of the cross-attention mechanism within a predetermined window/dilated window size. This approach not only excludes distant and irrelevant information but also reduces computational cost. Furthermore, we develop a lightweight Siamese backbone for extracting building and environmental features. Subsequently, we integrate an SWDCA module into this backbone, forming an efficient change detection network. Quantitative evaluations and visual analyses of thorough experiments verify that our method achieves top-tier accuracy on two building change detection datasets of remote sensing imagery, while also achieving a real-time inference speed of 33.2 FPS on a mobile GPU.

Effects of fast Fourier transform window size on the target strength spectra measurements of Japanese anchovy (Engraulis japonicus) using linear frequency modulated signals

Effects of fast Fourier transform window size on the target strength spectra measurements of Japanese anchovy (Engraulis japonicus) using linear frequency modulated signals

Cross-scale hierarchical spatio-temporal transformer for video enhancement

The diversity and complexity of degradations in low-quality videos pose non-trivial challenges on video enhancement to reconstruct the high-quality counterparts. Prevailing sliding window based methods represent poor performance due to the limitation of window size. Recurrent networks take advantage of long-term modeling to aggregate more information, resulting in significant performance improvements. However, most of them are trained on simple degraded data and can only tackle specific degradation. To break through the limitation, we propose a progressive alignment network, namely Cross-scale Hierarchical Spatio-Temporal Transformer (CHSTT), which leverages cross-scale tokenization to generate multi-scale visual tokens in the entire sequence to capture extensive long-range temporal dependencies. To enhance the spatial and temporal interactions, we introduce an innovative hierarchical Transformer, facilitating the computation of mutual multi-head attention across both spatial and temporal dimensions. Quantitative and qualitative assessments substantiate the superior performance of CHSTT compared to several state-of-the-art benchmarks across three distinct video enhancement tasks, including video super-resolution, video denoising, and video deblurring.

A Throughput Analysis Using a Non-Saturated Markov Chain Model for LTE-LAA and WLAN Coexistence

To address the severe spectrum shortage in mobile networks, the 3rd Generation Partnership Project (3GPP) standardized Long Term Evolution (LTE)-License Assisted Access (LAA) technology. The LTE-LAA system ensures efficient coexistence with other existing unlicensed systems by incorporating listen-before-talk functionality and conducting random backoff operations similar to those in the IEEE 802.11 distributed coordination function. In this paper, we propose an analytical model to calculate the throughput of each system in a scenario where a single LTE-LAA system shares an unlicensed channel with multiple wireless local area network (WLAN) systems. The LTE-LAA system is utilized for supplementary downlink transmission from the LTE-LAA eNodeB (eNB) to LTE-LAA devices. Our proposed analytical model uses a Markov chain to represent the random backoff operations of the LTE-LAA eNB and WLAN nodes under non-saturated traffic conditions and to calculate the impact of the clear channel assessment (CCA) performed by the LTE-LAA eNB. Through numerical results, we demonstrate how the throughput of both the LTE-LAA and WLAN systems is determined by the contention window size and CCA threshold of the LTE-LAA eNB.

A Channel Measurement-Based Listen-Before-Talk Algorithm for LTE-LAA and WLAN Coexistence

To support the coexistence of long-term evolution (LTE)-license-assisted access (LAA) and wireless local area network (WLAN) in unlicensed bands, the load-based listen-before-talk (LB-LBT) scheme has been developed, incorporating channel sensing and backoff functions similar to those used in WLAN. In the LB-LBT scheme, the contention window size and clear channel assessment (CCA) threshold of the LTE-LAA eNodeB (eNB) significantly influences its transmission probability and the interference caused by concurrent WLAN transmissions outside the CCA range. However, most existing LB-LBT schemes use fixed contention window sizes and CCA thresholds, irrespective of the channel congestion status. To address this limitation, in this paper, we propose a channel measurement-based LBT (CM-LBT) scheme to enhance overall system throughput while ensuring fairness between LTE-LAA and WLAN systems. Our proposed CM-LBT scheme adaptively adjusts the contention window size and CCA threshold of LTE-LAA eNB in the LB-LBT scheme, according to the current channel access activities of LTE-LAA and WLAN systems. Through simulations, we evaluate the performance of our proposed CM-LBT scheme against existing LBT schemes by assessing the throughput of LTE-LAA and WLAN systems, as well as the fairness between them, using a reward function.

Annual Cropping Intensity Dynamics in China from 2001 to 2023

Spatial and temporal information about cropping patterns of single and multiple crops is important for monitoring crop production and land-use intensity. We used time-series MODIS NDVI 8-day composite data to develop annual cropping pattern products at a 250 m spatial resolution for China, covering the period from 2001 to 2023. To address the potential impacts of varying parameters in both data pre-processing and the peak detection algorithm on the accuracy of cropping pattern mapping, we employed a grid-search method to fine-tune these parameters. This process focused on optimizing the Savitzky–Golay smoothing window size and the peak width parameters using a calibration dataset. The results highlighted that an optimal combination of a five to seven MODIS composite window size in Savitzky–Golay smoothing and a peak width of four MODIS composites achieved good overall mapping accuracy. Pixel-wise accuracy assessments were conducted for the selected mapping years of 2001, 2011, and 2021. Overall accuracies were between 89.7% and 92.0%, with F1 scores ranging from 0.921 to 0.943. Nationally, this study observed a fluctuating trend in multiple cropping percentages, with a notable increase after 2013, suggesting shifts toward more intensive agricultural practices in recent years. At a finer spatial scale, the combination of Mann–Kendall and Sen’s slope analyses revealed that approximately 12.9% of 3 km analytical windows exhibited significant changes in cropping intensity. We observed spatial clusters of increasing and decreasing crop intensity trends across provinces such as Hebei, Shandong, Shaanxi, and Gansu. This study underscores the importance of data smoothing and peak detection methods in analyzing high temporal resolution remote sensing data. The generation of annual single/multiple cropping pattern maps at a 250 m spatial resolution enhances our comprehension of agricultural dynamics through time and across different regions.

Impact of the STFT Window Size on Classification of Grain-Oriented Electrical Steels from Barkhausen Noise Time–Frequency Spectrograms via Deep CNNs

The Magnetic Barkhausen Noise (MBN) is a non-destructive testing method, which, due to its high sensitivity to changes in the microstructure of the material, is increasingly being applied with success as a tool for evaluation of magnetic material state and properties. However, it is no less difficult to analyze the measurement signals and their correct interpretation due to the complex, non-deterministic and stochastic nature of the Barkhausen phenomenon. Depending on the material to be examined, a signal with different characteristics can be observed. Frequently, a signal with multi-phase Barkhausen activity characteristics is obtained, like in the case of grain-oriented electrical steels. Due to the increased computational capabilities of computers, more and more advanced signal analysis methods are being used and artificial intelligence is being involved as well. Recently, the time–frequency (TF) approach for MBN signal analysis was introduced and discussed in several papers, where short-time Fourier Transform (STFT) found frequent application with promising results. Due to the automation of the search for diagnostic patterns, the stage of selecting transformation parameters becomes extremely important in the process of preparing training data for evaluation algorithms. This paper investigates the influence of the STFT computational window size on the material state evaluation results obtained using convolutional neural network (CNN). The studies were performed for MBN signals obtained from grain-oriented electrical steel with anisotropic properties. The carried out work made it possible to draw connections on the importance of the choice of the window during the implementation of CNN network training.

Shear Wave Travel Time Prediction using Well Log Filtering and Machine Learning

Shear wave travel time (also known as Delta-T Shear and commonly abbreviated as DTS) is an important parameter in petroleum for exploration, production, and characterization of borehole stability. Direct measurement of DTS is often limited by high costs and a constraint of geography, making machine learning (ML) predictive approaches necessary. This study aims to explore the effectiveness of ML models in predicting DTS, emphasizing the importance of data preprocessing techniques to improve prediction accuracy. Preprocessing techniques include Yeo-Johnson transformation to handle non-normality, outlier elimination using z-score, and data smoothing using the Savitzky-Golay filter and median filter. Incorporating smoothing techniques can fill important gaps in some existing studies and may improve the performance of machine learning models in predicting DTS, particularly in situations with limited or noisy data. Four ML models were tested in this study, namely Linear Regression (LR), K-Nearest Neighbors (KNN), Extreme Gradient Boosting (XGBoost), and Random Forest (RF), with performance evaluation based on metrics RMSE (Root Mean Squared Error), MAE (Mean Absolute Error), and R2 (coefficient of determination). The results showed that the RF model produced the best performance with RMSE of 9.41, MAE of 6.35, and R2 of 0.90 in scenarios with Yeo-Johnson transformation, outlier elimination, and smoothing techniques using a median filter with a window size of 5.

SymSwin: Multi-Scale-Aware Super-Resolution of Remote Sensing Images Based on Swin Transformers

Despite the successful applications of the remote sensing image in agriculture, meteorology, and geography, its relatively low spatial resolution is hindering the further applications. Super-resolution technology is introduced to conquer such a dilemma. It is a challenging task due to the variations in object size and textures in remote sensing images. To address that problem, we present SymSwin, a super-resolution model based on the Swin transformer aimed to capture a multi-scale context. The symmetric multi-scale window (SyMW) mechanism is proposed and integrated in the backbone, which is capable of perceiving features with various sizes. First, the SyMW mechanism is proposed to capture discriminative contextual features from multi-scale presentations using corresponding attentive window size. Subsequently, a cross-receptive field-adaptive attention (CRAA) module is introduced to model the relations among multi-scale contexts and to realize adaptive fusion. Furthermore, RS data exhibit poor spatial resolution, leading to insufficient visual information when merely spatial supervision is applied. Therefore, a U-shape wavelet transform (UWT) loss is proposed to facilitate the training process from the frequency domain. Extensive experiments demonstrate that our method achieves superior performance in both quantitative metrics and visual quality compared with existing algorithms.

Signal Reconstruction Based on Time‐Varying Sliding Window in WSNs Using Compressed Sensing

ABSTRACTThis paper presents a new algorithm that utilizes compressed sensing (CS) for reconstruction of wireless sensor networks (WSNs) data with spatial and temporal correlation. The proposed method utilizes a time‐varying sliding window mechanism that dynamically adjusts both the window size and the number of measurements. This flexibility allows the algorithm to exploit spatio‐temporal correlations effectively, ensuring that data within the window remains sparse and thus more compressible. By dynamically varying the number of measurements, the algorithm equitably distributes the sampling rate across different time slots, adapting to changes in signal characteristics and minimizing transmission costs. Simulation results demonstrate that our proposed algorithm outperforms other CS reconstruction methods by achieving higher reconstruction precision while requiring fewer transmissions. This is achieved through a decentralized data‐window framework that maximizes the use of prior signal information, leading to improved signal recovery performance in diverse WSN scenarios.

Fusion of Deep Learning-based and Spectral Features for Hyperspectral Image Analysis

Abstract. Currently, deep neural networks have become one of the most effective tools in computer vision. However, in the field of hyperspectral remote sensing image analysis, their practical application is limited, as it requires manual labeling of a large amount of data. Since this process is time-consuming and expensive, an attractive option is the use of pre-trained neural networks designed to work with color images. However, to take advantage of hyperspectral images, such neural networks must be equipped with some mechanism to take into account the detailed spectral information contained in such images.In this work, we propose to combine deep features computed using pre-trained convolutional neural networks (specifically Resnet18) with spectral features of hyperspectral images. The proposed scheme works on the basis of combining the selected type of distances (Euclidean distance, spectral angle, Hellinger divergence) in spectral and embedding spaces with the subsequent synthesis of features in a space of a given dimensionality. The proposed scheme does not require any training, except for the selection of several parameters (spatial window size, dimensionality of the synthesized space, fusion coefficient). Experiments conducted on known hyperspectral scenes (Indian Pines, Salinas, Pavia University, Kennedy Space Center) show the advantages of the proposed approach. The issue of train-test sample splitting is considered.

Real -Time Network Packet Classification Exploiting Computer Vision Architectures

The upcoming 6G and NextG networks underscore the necessity of sophisticated security methods based on Artificial Intelligence (AI) in order to detect malicious activity and adjust to new threats. Because computer vision techniques may be used to recognize complex patterns, their incorporation into the cybersecurity industry is a promising development. In this work, we present a computationally effective categorization technique that enforces the real-time conversion of packets into pictures by directly acting upon the raw packets gathered at base stations. The suggested solution's novel features include its lightweight implementation, which well satisfies the requirements of upcoming 6G networks, and its network edge operation, which permits early threat detection as near to the packet origin as feasible. We examine the efficacy of this methodology in terms of F1-score and prediction time by employing cutting-edge computer vision architectures and a customized Convolutional Neural Network (CNN) to tackle an intrusion detection task utilizing a substantial 5G dataset. The CNN design is superior than complicated models, as demonstrated by the results of experiments. The CNN consistently beats the other cutting-edge computer vision models over several packet window sizes N (i.e., 10, 50, and 100 packets), reaching very high F1-scores (0.99593, 0.99860, and 0.99895). A scalability investigation reveals a trade- off between the performance and scalability of CNN, with higher N values resulting in longer prediction times. However, the scalability of the other computer vision models is superior, allowing for an ideal model selection free of compromises