Search

The Acacia longifolia species is known for its rapid growth and dissemination, causing loss of biodiversity in the affected areas. In order to avoid the uncontrolled spread of this species, it is important to effectively monitor its distribution on the agroforestry regions. For this purpose, this paper proposes the use of Convolutional Neural Networks (CNN) for the detection of Acacia longifolia, from images acquired by an unmanned aerial vehicle. Two models based on the same CNN architecture were elaborated. One classifies image patches into one of nine possible classes, which are later converted into a binary model; this model presented an accuracy of 98.6 % and 98.5 % in the validation and training sets, respectively. The second model was trained directly for binary classification and showed an accuracy of 98.8 % and 98.7 % for the validation and test sets, respectively. The results show that the use of multiple classes, useful to provide the aerial vehicle with richer semantic information regarding the environment, does not hamper the accuracy of Acacia longifolia detection in the classifier’s primary task. The presented system also includes a method for increasing classification’s accuracy by consulting an expert to review the model’s predictions on an automatically selected sub-set of the samples.

Machine learning service API allows model owners to monetize proprietary models by offering prediction services to third-party users. However, existing literature shows that model parameters are vulnerable to extraction attacks which accumulate prediction queries and their responses to train a replica model. As countermeasures, researchers have proposed to reduce the rich API output, such as hiding the precise confidence. Nonetheless, even with response being only one bit, an adversary can still exploit fine-tuned queries with differential property to infer the decision boundary of the underlying model. In this paper, we propose boundary differential privacy (BDP) against such attacks by obfuscating the prediction responses with noises. BDP guarantees an adversary cannot learn the decision boundary of any two classes by a predefined precision no matter how many queries are issued to the prediction API. We first design a perturbation algorithm called boundary randomized response for a binary model. Then we prove it satisfies e-BDP, followed by a generalization of this algorithm to a multiclass model. Finally, we generalize a hard boundary to soft boundary and design an adaptive perturbation algorithm that can still work in the latter case. The effectiveness and high utility of our solution are verified by extensive experiments on both linear and non-linear models.

This paper presents an extension of the Binary Model Reference Adaptive Control (BMRAC) for uncertain multivariable (square) systems with non-uniform arbitrary relative degrees using only output feedback and its application to inertially stabilized platforms using a two degree of freedom gimbal as actuator. The BMRAC is a robust adaptive strategy with good transient performance, thus useful for uncertain systems, and the multivariable framework is suitable to deal with mechanical unbalances. Using a newly proposed differentiator with dynamic gains based on higher-order sliding mode, the proposed controller achieves global and exact tracking. To illustrate the effective of the proposed solution, simulations are presented using real-word data obtained from an instrumented vehicle in an irregular ground.

This study investigated the implications of board independence and foreign ownership on audit quality of manufacturing firms in Nigeria. The specific objectives of the study are to examine the effects of board independence as well as foreign ownership on audit quality of manufacturing firms quoted in Nigeria. Secondary data were carefully collected from a total of fifty eight (58) quoted manufacturing firms in Nigeria for the period (2010 – 2018) and the binary model of regression (logit, probit and gombit) was properly used for hypotheses testing. The outcome reveals that board independence had a positive and insignificant influence on audit quality while foreign ownership had a positive and significant influence on audit quality. The study therefore recommends that composition of the board should be such that its function is not undermined and one of such ways is to have an appropriate mixture with non-executive directors. Also having foreign ownership could enhance audit quality given the different corporate cultures they may possess.

International audience; We address the runway scheduling problem under consideration of winter operations. During snowfall, runways have to be temporarily closed in order to clear them from snow, ice and slush. We propose an integrated optimization model to simultaneously plan snow removal for multiple runways and to assign runways and takeoff and landing times to aircraft. For this winter runway scheduling problem, we present a time-discrete binary model formulation using clique inequalities and an equivalent constraint programming model. To solve the winter runway scheduling problem optimally, we propose an exact solution methodology. Our start heuristic based on constraint programming generates a feasible initial start solution. We use a column generation scheme, which we initialize with a heuristic solution, to identify all variables of the binary program which are required to solve it optimally. Finally, we apply a branch-and-bound procedure to our resulting binary program. Additionally, we present an enhanced time discretization method to balance model size and solution quality. We apply our algorithm to realistic instances from a large international airport. An analysis of resulting model sizes proves the ability of our approach to significantly reduce the number of required variables and constraints of the time-discrete binary program. We also show that our method computes optimal schedules in a short amount of time and often outperforms a time-continuous formulation as well as a pure constraint programming approach.

The current research used human-coded Reddit posts categorized by already established food parenting concepts (coercive control, structure, autonomy support, recipes) as a basis for machine learning models, with the objective of providing insight into topics related to feeding children discussed on social media and to provide a way for future research to use our trained machine-learned model. Reddit posts from specific, parenting-related subreddits were collected and labeled as they related to aspects of child-feeding behavior. Posts were then put through text pre-processing, converted into TF-IDF vectors, and used to train support vector machine binary and multiclass classification models. Other classifiers and text-preprocessing steps were also tested. After training, the binary model was able to classify posts with 86.1% accuracy as being about child feeding or not, up from a baseline accuracy of 57.6%. The multiclass model yielded a 79.1% accuracy to classify posts related to four categories of child feeding concepts (coercive control, autonomy support, structure, recipes), up from a baseline of 51.9%. The comparison models were found to perform less favorably. The best performing binary model is publicly available for use via the Social Media Macroscope and we provide details on how to use this model. Information is presented such that other researchers and professionals interested in examining issues related to feeding children posted on social media could effectively utilize the same approach.

Threshold method is an important image segmentation method, which has been widely used in image segmentation. For this method, it is very important to choose a good threshold. The traditional threshold segmentation algorithm is implemented by exhaustive method, which makes the solution efficiency very low. This paper presents a collective neurodynamic optimization algorithm to solve the problem of binary optimization in the image segmentation. The problem of image segmentation based on threshold is transformed into binary optimization with constraints. Then, a collective neurodynamic optimization algorithm is introduced which combined with feedback neural network and particle swarm optimization (PSO) algorithm. And the linear programming relaxation constraint method is used to relax binary constraints. It is proved by numerical simulation that the feedback neural network algorithm can converge to the exact local optimal solution of the model and the PSO algorithm can get a better local optimal solution. Finally, several sets of comparative experiments are presented. The feasibility of our proposed method is verified; the experimental results demonstrate the effectiveness of our approach in image segmentation. In this study, a collective neurodynamic optimization was proposed for the image segmentation problem. In the future, we expect that multiple centralized neurodynamic models and intelligent algorithms can be used to solve the problem and improve the convergence speed of the solved model.

This paper introduces a receiver who perceives ambiguity in a binary model of Bayesian persuasion. The sender has a well-defined prior, while the receiver considers an interval of priors and maximizes a convex combination of worst and best expected payoffs ( $$\alpha $$ -maxmin preferences). We characterize the sender’s optimal signal and find that the receiver’s payoff differences across states given each action (sensitivities), play a fundamental role in the characterization and the comparative statics. If the sender’s preferred action is the least (most) sensitive one, then the sender’s equilibrium payoff, as well as the sender’s preferred degree of receiver ambiguity, is increasing (decreasing) in the receiver’s pessimism. We document a tendency for ambiguity-sensitive receivers to be more difficult to persuade.

Planning for new urban–industrial communities in deserts is a main objective of the Egyptian government. Allocation of such communities should comply with both environmental protection and sustainable consumption of resources. The current study explores such approach, through mapping potential zones for new urban–industrial communities in Matrouh Governorate, Western Desert, Egypt. A spatial multidisciplinary approach has been adopted to facilitate zoning based on satellite data, spatial decision support systems and a geospatial model. Landsat 8, Shuttle Radar Topography Mission in addition to several thematic maps were input for the geospatial model. Three sub-models were created for three themes (visions), two of which were created using spatial multicriteria analysis (land resources and infrastructures), and the third sub-model, namely environmental protection, was created using a binary model. The analytical hierarchy process method was applied to assign relative weights for the resources theme, while the rank-sum method was used for the infrastructure theme. The three sub-models were aggregated to produce the suitability index for three zoning visions. Such index was classified and the highest three suitability values in each suitability index were extracted being the most optimum zones. Such zones are found in the northeastern part of the study area. The most optimum zones amounted to 23,863 km2, 47,707 km2 and 25,474 km2; equivalent to 14.8%, 29.7% and 15.9% for the first, second and third visions, respectively. The resultant maps can be used in site selection, allocation, decisions and for planning new urban communities.

Depending on the available experimental data (T, p, x, y, and HE), the vapor–liquid equilibrium (VLE) data of the completely miscible isobaric systems at low pressure are accurately correlated and ...

The dynamic change of urban thermal environment caused by the change of land use type has become one of the important problems of urban ecological environment protection. In Guilin city as research area, based on the Google Earth Engine (GEE), the random forest algorithm was used to classify the land use classification of Landsat remote sensing images in 2010, 2014 and 2018, and the mono-window algorithm was used to calculate the surface temperature. The surface vegetation was solved according to the NDVI pixel binary model. Coverage, and finally dynamic statistics and comparative analysis of land use, vegetation cover and surface temperature. The main results as follows. (1) From 2010 to 2018, the average temperature in the main urban area of Guilin is on the rise (increased by 1.29 °C), and the temperature zones in each class are converted from low temperature zone, lower temperature zone and medium temperature zone to higher temperature zone and high temperature zone. (2) Lower temperature zone and the low temperature zone is mainly distributed in vegetation and water body coverage areas, while the medium temperature zone, higher temperature zone and the high temperature zone are mainly distributed in construction land and unused land cover area. (3) High vegetation cover area in 2014–2018 (reduced by 31.34%) The main reason for the sharp decline is the substantial increase in the area of construction land (expansion 30.19%). (4) GEE-based random forest algorithm Land use classification had higher classification accuracy (more than 80% in all three periods). The results can provide scientific basis for improving urban thermal environment and scientific reference for the development strategy of Guilin city.

Rocky desertification is a common geo-ecological disasters in China are mainly distributed in southwest karst region, and a wide range of further deterioration. Based on the theory of decision tree Guangxi rocky information extraction, selection of experimental data of Guangxi Zhuang Autonomous Region in 2005 TM image. First of remote sensing images after geometric correction image registration and other pretreatment. Secondly based on binary model of pixel, the Guangxi Zhuang Autonomous Region NDVI values and vegetation cover and slope analysis combining the results of Guangxi Zhuang Autonomous Region, the use of decision tree classification of remote sensing images, and finally get different levels of Guangxi Zhuang Autonomous Region rocky area and spatial distribution. The experimental results showed that: 2005 Guangxi rocky area of about 22,000 km2, accounting for 9% of the total land area in Guangxi, accounting for 24.30% of the karst area the overall classification accuracy of 89.03%, Kappa coefficient was 0.8417. From the classification results and the accuracy evaluation shows that the use of the information extracted rocky achieve better results.

Aim: The purpose of this research was to use environmental variables for predicting the probability of Hemiscorpius lepturus existence in the provinces where situated in the west of the Zagros Mountains. Materials and Methods: In this study, 64 occurrence records of the H. lepturus were extracted from the published documents available in electronic databases. MaxEnt model was used for predicting the ecological niches of this species. Normalized difference vegetation index (NDVI) and 19 climatic variables were used as the environmental variables affecting the distribution of this scorpion. The Jackknife test in the model was used to indicate the importance of variables to predict the probability of the presence of the studied species. The logistic threshold that was evaluated using a logistic regression algorithm showed the converting of the probability model into a binary model. The model was evaluated by area under the curve (AUC). The probability presence map of this scorpion was then prepared in ArcGIS 10.5 Software. Results: The results of the analysis showed that the most important environmental factor on the distribution of H. lepturus was the maximum temperature of the warmest month (Bio5) with a contribution rate of 43% and permutation importance of 8%. The Jackknife test revealed that NDVI did not gain any value when it used independently in the model. The logistic threshold was reported 0.255 for the maximum test sensitivity plus specificity. The AUC of the model was 0.7698, shows an acceptable value for model validity. Overall the hot spots for this toxic scorpion seem to be in Khuzestan, Lorestan, and Ilam Provinces of the studied area. Conclusion: Regarding our findings, MaxEnt algorithm, in combination with geographic information system contributed to revealing the effects of environmental variables on the probability of H. lepturus presence in the west of Zagros Mountains. These visualized maps as a warning alarm can be helpful to policymakers for managing, controlling, and monitoring the scorpionism in high-risk areas.

The analysis of non-metallic inclusions is crucial for the assessment of steel properties. Scanning electron microscopy (SEM) coupled with energy dispersive spectroscopy (EDS) is one of the most prominent methods for inclusion analysis. This study utilizes the output generated from SEM/EDS analysis to predict inclusion types from BSE images. Prediction models were generated using two different algorithms, Random Forest (RF) and convolutional neural networks (CNN), for comparison. For each method, three separate models were developed. Starting with a simple binary model to differentiate between inclusions and non-inclusions, then developing to more complex four and five class models. For the 4-class model, inclusions were split into oxides, sulfides, and oxy-sulfides, in addition to the non-inclusion class. The 5-class model included specific types of inclusions only, namely alumina, calcium aluminates, calcium sulfides, complex calcium-manganese sulfides, and oxy-sulfide inclusions. CNN achieved better accuracy for the binary (92%) and 4-class (78%) models, compared to RF (binary 87%, 4-class 75%). For the 5-class model, the results were similar, 60% accuracy for RF and 59% for CNN.

In this manuscript we use an exactly solvable stochastic binary model for regulation of gene expression to analyse the dynamics of response to a treatment aiming to modulate the number of transcripts of RKIP gene. We demonstrate the usefulness of our method simulating three treatment scenarios aiming to reestablish RKIP gene expression dynamics towards pre-cancerous state: i. to increase the promoter’s ON state duration; ii. to increase the mRNAs’ synthesis rate; iii. to increase both rates. We show that the pre-treatment kinetic rates of ON and OFF promoter switching speeds and mRNA synthesis and degradation will affect the heterogeneity and time for treatment response. Hence, we present a strategy for reducing drug dosage by simultaneously targeting multiple kinetic rates. That enables a reduction of treatment response time and heterogeneity which in principle diminishes the chances of emergence of resistance to treatment. This approach may be useful for inferring kinetic constants related to expression of antimetastatic genes or oncogenes and on the design of multi-drug therapeutic strategies targeting master regulatory genes.

The high computational complexity and memory consumption of the deep Convolutional Neural Networks (CNNs) restrict their deployability in resource-limited embedded devices. To address this challenge, emerging solutions are proposed for neural network quantization and compression. Among them, Binary Neural Networks (BNNs) show their potential in reducing computational and memory complexity; however, they suffer from considerable performance degradation. One of the major causes is their non-differentiable discrete quantization implemented using a fixed sign function, which leads to output distribution distortion. In this paper, instead of using the fixed and naive sign function, we propose a novel adaptive Neural Encoder (NE), which learns to quantize the full-precision weights as binary values. Inspired by the research of neural network distillation, a distribution loss is introduced as a regularizer to minimize the Kullback-Leibler divergence between the outputs of the full-precision model and the encoded binary model. With an end-to-end backpropagation training process, the adaptive neural encoder, along with the binary convolutional neural network, could reach convergence iteratively. Comprehensive experiments with different network structures and datasets show that the proposed method can improve the performance of the baselines and also outperform many state-of-the-art approaches. The source code of the proposed method is publicly available at https://github.com/CQUlearningsystemgroup/LearningToBinarize.

We consider the class of linear predictors over all logical conjunctions of binary attributes, which we refer to as the class of combinatorial binary models (CBMs) in this paper. CBMs are of high knowledge interpretability but naïve learning of them from labeled data requires exponentially high computational cost with respect to the length of the conjunctions. On the other hand, in the case of large-scale datasets, long conjunctions are effective for learning predictors. To overcome this computational difficulty, we propose an algorithm,GRAfting for Binary datasets (GRAB), which efficiently learns CBMs within the$$L_1$$L1-regularized loss minimization framework. The key idea of GRAB is to adopt weighted frequent itemset mining for the most time-consuming step in the grafting algorithm, which is designed to solve large-scale$$L_1$$L1-RERM problems by an iterative approach. Furthermore, we experimentally showed that linear predictors of CBMs are effective in terms of prediction accuracy and knowledge discovery.

Recently, deep neural networks (DNNs) significantly outperform Gaussian mixture models in acoustic modeling for speech recognition. However, the substantial increase in computational load during the inference stage makes deep models difficult to directly deploy on low-power embedded devices. To alleviate this issue, structure sparseness and low precision fixed-point quantization have been applied widely. In this work, binary neural networks for speech recognition are developed to reduce the computational cost during the inference stage. A fast implementation of binary matrix multiplication is introduced. On modern central processing unit (CPU) and graphics processing unit (GPU) architectures, a 5–7 times speedup compared with full precision floatingpoint matrix multiplication can be achieved in real applications. Several kinds of binary neural networks and related model optimization algorithms are developed for large vocabulary continuous speech recognition acoustic modeling. In addition, to improve the accuracy of binary models, knowledge distillation from the normal full precision floating-point model to the compressed binary model is explored. Experiments on the standard Switchboard speech recognition task show that the proposed binary neural networks can deliver 3–4 times speedup over the normal full precision deep models. With the knowledge distillation from the normal floating-point models, the binary DNNs or binary convolutional neural networks (CNNs) can restrict the word error rate (WER) degradation to within 15.0%, compared to the normal full precision floating-point DNNs or CNNs, respectively. Particularly for the binary CNN with binarization only on the convolutional layers, the WER degradation is very small and is almost negligible with the proposed approach.

Aim and Objective: A metabolic pathway is an important type of biological pathway, which is composed of a series of chemical reactions. It provides essential molecules and energies for living organisms. To date, several metabolic pathways have been uncovered. However, their completeness is still on the way. A number of prediction methods have been built to assign chemicals into certain metabolic pathway, which can further be used to predict novel latent chemicals for a given metabolic pathway. However, they did not make use of chemical properties in a system level to construct prediction models. Method: In this study, we applied a network integration method, which can extract topological features from different chemical networks, representing chemical associations from their different properties, and fused several high-dimension vector representations into a low-dimension vector representation for each chemical. The compact vector representations were fed into the Support Vector Machine (SVM) to construct the prediction model. To tackle the problem that one chemical can participate in more than one pathway type, we construct an SVM-based binary prediction model for each pathway type to determine whether a given chemical can participate in the pathway type. Furthermore, the Synthetic Minority Over-sampling Technique (SMOTE) was adopted to weaken the influence of imbalanced dataset. Results and Conclusion: Each binary model gave a quite good performance and was superior to the classic prediction model, indicating that the proposed models can be useful tools for integrating heterogeneous information to assign chemicals into correct metabolic pathways.

Microbial symbionts are ubiquitous and can have significant impact on hosts. These impacts can vary in the sign (positive or negative) and degree depending on the identity of the interacting partners. Studies on host-symbiont associations indicate that subspecies (strain) genetic variation can influence interaction outcomes, making it necessary to go beyond species-level distinction to understand host-symbiont dynamics. In this review, we discuss examples of strain specificity found in host-symbiont associations, from binary model systems to the human microbiome. Although host and bacterial factors identified as mediators for specificity could be distinct at the molecular level, they generally fall into two broad functional categories: (1) those that contribute a required activity in support of the association and (2) those involved in antagonistic interactions with organisms outside of the association. We argue here based on current literature that factors from these two categories can work in concert to drive strain specificity and that this strain specificity must be considered to fully understand the molecular and ecological dynamics of host-symbiont associations, including the human microbiome.

The aim of this paper is to introduce an improved strategy for controlling COVID-19 and other pandemic episodes as an environmental disinfection culture for public places. The scheduling aims at achieving the best utilization of the available working day-time hours, which is calculated as the total consumed disinfection times minus the total loosed transportation times. The proposed problem in network optimization identifies a disinfection group who is likely to select a route to reach a subset of predetermined public places to be regularly disinfected with the most utilization of the available day-working hours. A Nonlinear Binary Model is introduced with a detailed real application case study involving improving the scheduling of Coronavirus disinfection process for some Educational Institutions as an example of crowded public places in Cairo, Egypt. The case study mathematical model is solved using a novel Discrete Binary Gaining-Sharing knowledge-based optimization algorithm (DBGSK). The results of this study show that the novel optimization algorithm can efficiently solve the proposed Problem.

This paper presents a comparative study on the performance of binary- and multi-class Deep Neural Network classification models that have been trained with the optimized hyperparameters. Four sequential models are developed using Tensorflow and Keras to perform sentiment classification of product reviews posted on Amazon, IMDb, Twitter and Yelp. Preprocessing is carried out to cleanse text reviews and to extract informative features. 5-fold cross evaluation results demonstrate that the final multiclass model perform with an accuracy of 74.7% and 72.4%, whereas the binary model achieves of 80.0% and 69.5% indicating better performance. The conclusion of this study is that an optimized binary model architecture can be used to train multiclass classification models and save significant amounts of computing time.

In this technological world encoding the images is one of the important activities to avoid monitor or hack by the third parties. Data protection becomes essential because the communication occurs frequently over open networks. This paper focuses on Binary model encryption using a form of stream ciphering. Due to the development of technology conversion of images into encoded format is more essential in any type of network. It helps to protect our data from the hacker who try to steel our information’s. Confidentially is particularly important for safe dissemination of the media over IP networks. While compression of image reduces the bandwidth, it is not safe to transmit compressed images alone. It is equally important to encrypt the images until they are transmitted to a network. Investigation shows that the methodology suggested is generating better results than the current procedures.

Over the past decade, high-utility itemset mining (HUIM) has received widespread attention that can emphasize more critical information than was previously possible using frequent itemset mining (FIM). Unfortunately, HUIM is very similar to FIM since the methodology determines itemsets using a binary model based on a pre-defined minimum utility threshold. Additionally, most previous works only focused on single, small datasets in HUIM, which is not realistic to any real-world scenarios today containing big data environments. In this work, the fuzzy-set theory and a MapReduce framework are both utilized to design a novel high fuzzy utility pattern mining algorithm to resolve the above issues. Fuzzy-set theory is first involved and a new algorithm called efficient high fuzzy utility itemset mining (EFUPM) is designed to discover high fuzzy utility patterns from a single machine. Two upper-bounds are then estimated to allow early pruning of unpromising candidates in the search space. To handle the large-scale of big datasets, a Hadoop-based high fuzzy utility pattern mining (HFUPM) algorithm is then developed to discover high fuzzy utility patterns based on the Hadoop framework. Experimental results clearly show that the proposed algorithms perform strongly to mine the required high fuzzy utility patterns whether in a single machine or a large-scale environment compared to the current state-of-the-art approaches.

In this paper, a sequential decomposition method to model the travel behavior of individuals in a mixed transportation network (transit and roads) is proposed. A mixed binary model is developed assuming the mode choice of the individuals depend on the accessibility level for different modes and parking fees of the vehicles at different locations, where the parking fees vary depending on the demand level and time of day. The efficiency of the proposed model is tested using simulated data. It is shown that by slicing the activity patterns and parallelizing some segments of the algorithm, the computation time increases linearly as the population size increases.

An improved scheduling the disinfection process of the new coronavirus (COVID-19) is introduced. The scheduling aims at achieving the best utilization of the available day time, which is calculated as the total disinfection time minus the total loss travelling time. In this regard, a new application problem is presented, which is called a travelling disinfection-man problem (TDP). The new problem (TDP) in network optimization resemble somehow the famous travelling salesman problems (TSP) but with basic distinct variations where a disinfection group is likely to select a route to reach a subset of predetermined places to be disinfected with the most utilization of the available day working hours. A nonlinear binary model is introduced with a detailed real application case study involving the improving the scheduling of coronavirus disinfection process for five contaminated faculties in Ain Shams University in Cairo, and the case study is solved using a novel discrete binary gaining-sharing knowledge-based optimization algorithm (DBGSK).

Based on Presidential Regulation Number 131 of 2015 on Determine of Underdeveloped Regions in 2015-2019, Maluku ranks 4th out of 23 provinces. Maluku has 11 regions, 8 of them are classified as underdeveloped regions. Classification of underdeveloped regions can be done using statistical analysis, namely the Binary Multivariate Adaptive Regression Spline (MARS). So, specific objectives to be achieved in this study are to determine the best Binary MARS model for classification and to calculate the accuracy of the Binary MARS model for the classification of underdeveloped regions in Maluku. After obtaining the classification results, we find out of GCV value for the MARS Binary model was 0.155 and the R2 value is 0.897. This model provided 100% accuracy in classifying the underdeveloped regions in Maluku.

In this paper, we propose variants of deep learning methods to segment head and operculum of the zebrafish larvae in microscopy images. In the first approach, we used a three-class model to jointly segment head and operculum area of zebrafish larvae from background. In the second, two-step, approach, we first trained binary segmentation model to segment head area from the background followed by another binary model to segment the operculum area within cropped head area thereby minimizing the class imbalance problem. Both of our approaches use a modified, simpler, U-Net architecture, and we also evaluate different loss functions to tackle the class imbalance problem. We systematically compare all these variants using various performance metrics. Data and open-source code are available at https://uliege.cytomine.org.

Rib fractures are injuries commonly assessed in trauma wards. Deep learning has demonstrated state-of-the-art accuracy for a variety of tasks, including image classification. This paper assesses the speed-accuracy trade-offs and general suitability of four popular convolutional neural networks to classify rib fractures from axial computed tomography imagery. We transfer learned InceptionV3, ResNet50, MobileNetV2, and VGG16 models, additionally training “decomposed” models comprised of taking only the first n blocks for each block for each architecture. Given that acute (new) fractures are generally most important to detect, we trained two types of models: a classful model with classes acute, old (healed), and normal (non-fractured); and a binary model with acute vs. the other classes. We found that the first 7 blocks of InceptionV3 achieved the best results and general speed-accuracy trade-off. The classful model achieved a 5-fold cross-validation average accuracy and macro recall of 96.00% and 94.0%, respectively. The binary model achieved a 5-fold cross-validation average accuracy, macro recall, and area under receiver operator characteristic curve of 97.76%, 94.6%, and 94.7%, respectively. On a Windows 10 PC with 32GB RAM and an Nvidia 1080ti GPU, the model's average CPU and GPU per-crop inference times were 13.6 and 12.2 ms, respectively. Compared to the InceptionV3 Block 7 classful model, a radiologist with 9 years of experience was less accurate but more sensitive to acute fractures; meanwhile, the deep learning model had fewer false positive diagnoses and better sensitivity to old fractures and normal ribs. The Cohen's Kappa between the two was 0.813.

Deep learning methods have already enjoyed an unprecedented success in medical imaging problems. Similar success has been evidenced when it comes to the detection of COVID-19 from medical images, therefore deep learning approaches are considered good candidates for detecting this disease, in collaboration with radiologists and/or physicians. In this paper, we propose a new approach to detect COVID-19 via exploiting a conditional generative adversarial network to generate synthetic images for augmenting the limited amount of data available. Additionally, we propose two deep learning models following a lightweight architecture, commensurating with the overall amount of data available. Our experiments focused on both binary classification for COVID-19 vs Normal cases and multi-classification that includes a third class for bacterial pneumonia. Our models achieved a competitive performance compared to other studies in literature and also a ResNet8 model. Our best performing binary model achieved 98.7% accuracy, 100% sensitivity and 98.3% specificity, while our three-class model achieved 98.3% accuracy, 99.3% sensitivity and 98.1% specificity. Moreover, via adopting a testing protocol proposed in literature, our models proved to be more robust and reliable in COVID-19 detection than a baseline ResNet8, making them good candidates for detecting COVID-19 from posteroanterior chest X-ray images., Highlights • Lightweight deep neural networks can accurately detect Covid-19, Bacterial Pneumonia and Normal cases from chest X-rays. • A Generative Adversarial Network has been developed that can provide good quality synthetic COVID-19 images. • Proposed models demonstrated robustness against adversarial inputs across binary and multi-class cases. • A layer of interpretability has been added to the models highlighting the areas that contributed to the detection decision.

We show that an equal-mass, temporary binary companion to the Sun in the solar birth cluster at a separation of $\sim 10^3 \; \mathrm{\; AU}$ would have increased the likelihood of forming the observed population of outer Oort cloud objects and of capturing Planet Nine. In particular, the discovery of a captured origin for Planet Nine would favor our binary model by an order of magnitude relative to a lone stellar history. Our model predicts an overabundance of dwarf planets, discoverable by LSST, with similar orbits to Planet Nine, which would result from capture by the stellar binary., 5 pages, 2 figures; accepted to ApJL

This study proposes and evaluates a diagnostic classification model framework for multiple-choice items. Models in the proposed framework have a two-level nested structure which allows for binary scoring (for correctness) and polytomous scoring (for distractors) at the same time. One advantage of these models is that they can provide distractor information while maintaining the statistical properties of the correct response option. We evaluated parameter recovery through a simulation study using Hamiltonian Monte Carlo algorithms in Stan. We also discussed three approaches to implementing the proposed modelling framework for different purposes and testing scenarios. We illustrated those approaches and compared them with a binary model and a traditional nominal model through an operational study.

Typically, binary classification lens-finding schemes are used to discriminate between lens candidates and non-lenses. However, these models often suffer from substantial false-positive classifications. Such false positives frequently occur due to images containing objects such as crowded sources, galaxies with arms, and also images with a central source and smaller surrounding sources. Therefore, a model might confuse the stated circumstances with an Einstein ring. It has been proposed that by allowing such commonly misclassified image types to constitute their own classes, machine learning models will more easily be able to learn the difference between images that contain real lenses, and images that contain lens imposters. Using Hubble Space Telescope (HST) images, in the F814W filter, we compare the usage of binary and multi-class classification models applied to the lens finding task. From our findings, we conclude there is not a significant benefit to using the multi-class model over a binary model. We will also present the results of a simple lens search using a multi-class machine learning model, and potential new lens candidates., PASP accepted, 14 pages, 10 figures, 4 tables

Statement of problem Tooth extraction therapy serves as a key initial step in many prosthodontic treatment plans. Dentists must make an appropriate decision on the tooth extraction therapy considering multiple determinants and whether a clinical decision support (CDS) model might help. Purpose The purpose of this retrospective records study was to construct a CDS model to predict tooth extraction therapy in clinical situations by using electronic dental records (EDRs). Material and methods The cohort involved 4135 deidentified EDRs of 3559 patients from the database of a prosthodontics department. Knowledge-based algorithms were first proposed to convert raw data from EDRs into structured data for feature extraction. Redundant features were filtered by a recursive feature-elimination method. The tooth extraction problem was then modeled alternatively as a binary or triple classification problem to be solved by 5 machine learning algorithms. Five machine learning algorithms within each model were compared, as well as the efficiency between 2 models. In addition, the proposed CDS was verified by 2 prosthodontists. Results The triple classification model outperformed the binary model with the F1 score of the Extreme Gradient Boost (XGBoost) algorithm as 0.856 and 0.847, respectively. The XGBoost outperformed the other 4 algorithms. The accuracy, precision, and recall of the XGBoost algorithm were 0.962, 0.865, and 0.830 in the binary classification and 0.924, 0.879, and 0.836 in the triple classification, respectively. The performance of the 2 prosthodontists was inferior to the models. Conclusions The CDS model for tooth extraction therapy achieved high performance in terms of decision-making derived from EDRs.

This introductory chapter describes the gender-diverse community, distinguishing between those who transition within the binary model of gender and those who identify as being neither female nor male, but somewhere along the continuum between these two poles or completely outside of the binary (i.e. “agender”). This introduction also provides an overview of the other chapters in the book, linking the issue of gender identity to the notion of religious liberty and previewing the conflict between the two.

Previously, the authors proposed a formalization of renewable resources rational use problem based on the representation of controlled system as a discrete dynamical system. In the particular case of structured ecosystem described by Leslie’s binary model, despite its non-linearity, it turned out that all optimal controls preserving this system belong to the certain hyperplane. This paper explores the conditions under which the positive boundary of a feasible set of problem with so-called quasi-preserving controls also contain a part of some hyperplane. In the process, we used a generalization of classical concept of map irreducibility—the concept of local irreducibility. Statements about the influence of the irreducibility property of discrete dynamical system step operator on the properties of an feasible set positive boundary are proved.

Binary Neural Networks (BNNs), known to be one among the effectively compact network architectures, have achieved great outcomes in the visual tasks. Designing efficient binary architectures is not trivial due to the binary nature of the network. In this paper, we propose a use of evolutionary search to facilitate the construction and training scheme when binarizing MobileNet, a compact network with separable depth-wise convolution. Inspired by one-shot architecture search frameworks, we manipulate the idea of group convolution to design efficient 1-Bit Convolutional Neural Networks (CNNs), assuming an approximately optimal trade-off between computational cost and model accuracy. Our objective is to come up with a tiny yet efficient binary neural architecture by exploring the best candidates of the group convolution while optimizing the model performance in terms of complexity and latency. The approach is threefold. First, we train strong baseline binary networks with a wide range of random group combinations at each convolutional layer. This set-up gives the binary neural networks a capability of preserving essential information through layers. Second, to find a good set of hyperparameters for group convolutions we make use of the evolutionary search which leverages the exploration of efficient 1-bit models. Lastly, these binary models are trained from scratch in a usual manner to achieve the final binary model. Various experiments on ImageNet are conducted to show that following our construction guideline, the final model achieves 60.09% Top-1 accuracy and outperforms the state-of-the-art CI-BCNN with the same computational cost., Comment: Accepted by CVPR2020

We consider selection of random predictors for a high-dimensional regression problem with a binary response for a general loss function. An important special case is when the binary model is semi-parametric and the response function is misspecified under a parametric model fit. When the true response coincides with a postulated parametric response for a certain value of parameter, we obtain a common framework for parametric inference. Both cases of correct specification and misspecification are covered in this contribution. Variable selection for such a scenario aims at recovering the support of the minimizer of the associated risk with large probability. We propose a two-step selection Screening-Selection (SS) procedure which consists of screening and ordering predictors by Lasso method and then selecting the subset of predictors which minimizes the Generalized Information Criterion for the corresponding nested family of models. We prove consistency of the proposed selection method under conditions that allow for a much larger number of predictors than the number of observations. For the semi-parametric case when distribution of random predictors satisfies linear regressions condition, the true and the estimated parameters are collinear and their common support can be consistently identified. This partly explains robustness of selection procedures to the response function misspecification.

Recent methods have significantly reduced the performance degradation of Binary Neural Networks (BNNs), but guaranteeing the effective and efficient training of BNNs is an unsolved problem. The main reason is that the estimated gradients produced by the Straight-Through-Estimator (STE) mismatches with the gradients of the real derivatives. In this paper, we provide an explicit convex optimization example where training the BNNs with the traditionally adaptive optimization methods still faces the risk of non-convergence, and identify that constraining the range of gradients is critical for optimizing the deep binary model to avoid highly suboptimal solutions. For solving above issues, we propose a BAMSProd algorithm with a key observation that the convergence property of optimizing deep binary model is strongly related to the quantization errors. In brief, it employs an adaptive range constraint via an errors measurement for smoothing the gradients transition while follows the exponential moving strategy from AMSGrad to avoid errors accumulation during the optimization. The experiments verify the corollary of theoretical convergence analysis, and further demonstrate that our optimization method can speed up the convergence about 1:2x and boost the performance of BNNs to a significant level than the specific binary optimizer about 3:7%, even in a highly non-convex optimization problem., Comment: 10 pages, 4 figures, 2 tables

Model management is a central activity in Software Engineering. The most challenging aspect of model management is to keep models consistent with each other while they evolve. As a consequence, there has been increasing activity in this area, which has produced a number of approaches to address this synchronization challenge. The majority of these approaches, however, is limited to a binary setting; i.e. the synchronization of exactly two models with each other. A recent Dagstuhl seminar on multidirectional transformations made it clear that there is a need for further investigations in the domain of general multiple model synchronization simply because not every multiary consistency relation can be factored into binary ones. However, with the help of an auxiliary artifact, which provides a global view over all models, multiary synchronization can be achieved by existing binary model synchronization means. In this paper, we propose a novel comprehensive system construction to produce such an artifact using the same underlying base modelling language as the one used to define the models. Our approach is based on the definition of partial commonalities among a set of aligned models. Comprehensive systems can be shown to generalize the underlying categories of graph diagrams and triple graph grammars and can efficiently be implemented in existing tools.

The rapid development of large pre-trained language models has greatly increased the demand for model compression techniques, among which quantization is a popular solution. In this paper, we propose BinaryBERT, which pushes BERT quantization to the limit by weight binarization. We find that a binary BERT is hard to be trained directly than a ternary counterpart due to its complex and irregular loss landscape. Therefore, we propose ternary weight splitting, which initializes BinaryBERT by equivalently splitting from a half-sized ternary network. The binary model thus inherits the good performance of the ternary one, and can be further enhanced by fine-tuning the new architecture after splitting. Empirical results show that our BinaryBERT has only a slight performance drop compared with the full-precision model while being 24x smaller, achieving the state-of-the-art compression results on the GLUE and SQuAD benchmarks. Code will be released.

This study extends a 1-parameter Z-Z-Z binary diffusion coefficient model to ternary and multicomponent systems. The cross-binary and ternary interaction parameters in atomic mobility (diffusion coefficient) assessments were systematically tested on 4 ternary solid solutions: fcc Ag-Au-Cu, fcc Co-Fe-Ni, fcc Cu-Fe-Ni, and bcc Nb-Ti-V. A simple combination of the Z-Z-Z binary model parameters without any additional fitting parameters already provides impressive predictions of the ternary diffusion coefficients when the 3 cross-binary parameters are set to be the corresponding binary Z-Z-Z model interaction parameters, leading to a robust Z-Z-ternary diffusion model. Employment of 3 independent cross-binary fitting parameters leads to an even better binary and cross-binary parameters only (BCBPO) model. Recommendations are rendered based upon the amount of available experimental ternary diffusion coefficients. These recommendations will substantially reduce the number of fitting parameters and improve the robustness of the resultant atomic mobility and diffusion coefficient databases for computational materials design.

Entry railings at escalator/stairs system are intervention measures for passenger flow and play a crucial role in local infrastructure optimization in mass transit stations. This paper presents an analysis of the choice between escalators and stairs with mixed logit models, by considering for the first time the length of the entry railing as one of the crucial parameters. Based on a pilot questionnaire survey, a large-scale field study was conducted in Changsha, China and a sample of 11010 passengers was drawn. The datasets of the whole sample and the grouped samples were applied for model calibrations, respectively. The results indicate that the sensitivity to the railing length show significant heterogeneity in preferences, as well as middle age, luggage carrying and density in front of the escalator (ESDEN). All models are validated to have good prediction accuracy. The mixed model outperforms the binary model in predicting the stairs usage, but slightly overestimates the escalator usage. The analysis reinforces earlier findings showing that lengthening the railing (2.5 ∼ 3.9 m) reduces the sensitivity of individual diversity to dynamic ESDEN, thus further enhancing the predictive ability by 4.35%. Our results provide unique insights for planners and policymakers on designing and managing transportation systems.

As the dry bulk shipping market seems to have been stuck in a trough period for a long time, investors need to pay more attention to their investment strategies to survive during this period. This study aimed to find a suitable model to assess dry bulk ship investment decisions in the tough and peak periods based on real options theories. Two options, involving an abandonment option and a deferrable option, were used to define investors’ responses to the uncertainty in investment processes such as stopping or selling vessels. The option valuation was solved by using a binomial valuation model, due to data limitations. In accordance with shipping cycle theories, different volatility parameters for the tough and peak periods were calculated using a generalized autoregressive conditional heteroskedasticity (GARCH) model. The application of the real options model to a case study involving secondhand ship trading indicated its viability. According to the results of the case study, the new model has adv...