Search

Boosting, as a meta-algorithm for ensemble learning, have been widely applied to variety popular machine learning algorithms. However, noises in training and testing datasets could significantly affect the performance of boosting algorithm. SAMME pays too much attention to samples that are not correctly classified in multiple iterations. These samples could be mislabeled samples that cannot be correctly classified, so the classifier cannot learn the actual distribution of the original data. To solve this problem, in this paper, we proposed a rectified algorithm R.SAMME based on multi-class classification algorithm SAMME by limiting the weight of each sample based on current accuracy. We evaluate our approach on UCI benchmark datasets, experiments show that R.SAMME has better performance in noisy datasets.

The Multi-Objective Evolutionary Algorithm based on Decomposition (MOEA/D) is a popular algorithm for solving Multi-Objective Problems (MOPs). The main component of MOEA/D is to decompose a MOP into easier sub-problems using a set of weight vectors. The choice of the number of weight vectors significantly impacts the performance of MOEA/D. However, the right choice for this number varies, given different MOPs and search stages. We adaptively change the number of vectors by removing unnecessary vectors and adding new ones in empty areas of the objective space. Our MOEA/D variant uses the Consolidation Ratio to decide when to change the number of vectors and to decide where to add or remove these weighted vectors. We investigate the effects of this adaptive MOEA/D against MOEA/D with a poorly chosen set of vectors, a MOEA/D with fine-tuned vectors and MOEA/D with Adaptive Weight Adjustment on two commonly used benchmark functions. We analyse the algorithms in terms of hypervolume, IGD and entropy performance. Our results show that the proposed method is equivalent to MOEA/D with fine-tuned vectors and superior to MOEA/D with poorly defined vectors. Thus, our adaptive mechanism mitigates problems related to the choice of the number of weight vectors in MOEA/D, increasing the final performance of MOEA/D by filling empty areas of the objective space and avoiding premature stagnation of the search progress.

In recent years, many different areas of research have utilized neural networks (NNs). Many have investigated weights in NNs as well as structural optimization of NNs via neural architecture search. In this paper, we apply weight training during neural architecture search by Ant Colony Optimization(ACO) to two problems from OpenAI Gym and one problem from pybullet-gym controlled by NNs. We also compare the timing of when the weight training is performed, which is before the architecture search, during architecture search and after architecture search. It was found that performing architecture search by ACO and weight training simultaneously is effective for increasing the score of NNs and that by performing weight training before or at the same time as architecture search, the score was increased statistically significantly for all problems compared with fully-connected NN and the score by performing weight training after architecture search was increased statistically significantly for only one problem from pybullet-gym.

Particle swarm optimization (PSO) is a new evolutionary algorithm developed in recent years. It is easy to implement with few parameters, but it is often easy to fall into local optimization in the later period. This paper introduces the interbreeding algorithm in genetic algorithms (GA) to increase population size diversity, and uses the inertia weight adjustment method of the chaos algorithm mechanism to adjust the inertia factor, and proposes an adaptive interbreeding hybrid particle swarm optimization (AIHPSO). In this paper, six representative nonlinear experimental functions are used to simulate and compare the algorithms. The results prove that AIHPSO plays a better role in a complex optimization process. It can improve local development capabilities, enhance convergence speed and the accuracy is significantly improved, while avoiding the problems of premature maturity and local optimization.

There are various of skin pigmented lesions with high risk. Melanoma is one of the most dangerous forms of skin cancer. It is one of the important research directions of medical artificial intelligence to carry out classification research of skin pigmented lesions based on deep learning. It can assist doctors to make clinical diagnosis and make patients receive treatment as soon as possible to improve survival rate. Aiming at the similar and imbalanced dermoscopic image data of pigmented lesions, this paper proposes a deep residual network improved by Squeeze-and-Excitation module, and dynamic update class-weight, in batches, with model ensemble adjustment strategies to change the attention of imbalanced data. The results show that the above method can increase the average precision by 9.1%, the average recall by 15.3%, and the average F1-score by 12.2%, compared with the multi-class classification using the deep residual network. Thus, the above method is a better classification model and weight adjustment strategy.

This study aims to identify the type of coffee powder aroma from the coffee beans blending using backpropagation artificial neural network (ANN). Backpropagation is a controlled training implementing a weight adjustment pattern to achieve a minimum error value between the the predicted and the actual output. Within this study, the coffee aroma testing utilized electronic tasting sensor system consisted of 4 sensors namely TGS 2611, TGS 2620, TGS 2610 and TGS 2602. The coffee aroma monitoring and data collection in this system applied LabVIEW software as a virtual instrumentation. The testing result of this ANN was able to distinguish the coffee variety of Robusta, Arabica coffee powder and the one without any coffee aroma. The backpropagation architecture was formed by 3 layers consisting of 1 input layer with 4 input nerve cells, 1 hidden layer with 8 neural cells, and 2 output layers by applying the backpropagation training algorithm. The training data was taken from 70 data samples of each circumstance of coffee with 5 testing times. The results of the training and testing showed that the established backpropagation was capable to identify and differenciate the coffee powder in accordance with the given input with different average success rate; 91.96% for Robusta coffee, 100 % for Arabica coffee, and no 84.24% for without coffee aroma.

Aiming at the problems of obstacle avoidance and bullet avoidance during the patrol swarm, this paper analyzed the defects of the classical artificial potential field, proposed an adjustable escape method, which establish the relationship between the adjustment coefficient and the distance. This method avoid too large or too small escape force that get the bullet into new local shock problem near the target. Then given the weight calculation and parameter selection method, restricted the escape motion by kinematics according to the constraints in the actual motion. This improved method can effecting solve the problem of avoidance in dynamic and complex environment.

Extreme learning machines (ELMs) are an interesting alternative to multilayer perceptrons because ELMs, in practice, require the optimization only of the number of hidden neurons by grid-search with cross-validation. Nevertheless, a large number of hidden neurons obtained after training is a significant drawback for the effective usage of such classifiers. In order to overcome this drawback, we propose a new approach to prune hidden layer neurons and achieve a fixed-size hidden layer without affecting the classifier’s accuracy and even improving it. The main idea is to leave out non-relevant or very similar neurons to others that are not necessary to achieve a simplified, yet accurate, model. In our proposal, differently from previous works based on genetic algorithms and simulated annealing, we choose only a subset of those neurons generated at random to belong to the hidden layer without weight adjustment or increasing of the hidden neurons. In this work, we compare our proposal called simulated annealing for pruned ELM (SAP-ELM) with pruning methods named optimally pruned ELM, genetic algorithms for pruning ELM and sparse Bayesian ELM. On the basis of our experiments, we can state that SAP-ELM is a promising alternative for classification tasks. We highlight that as our proposal achieves fixed-size models, it can help to solve problems where the memory consuming is crucial, such as embeded systems, and helps to control the maximum size that the models must reach after the training process.

Multiobjective evolutionary algorithm based on decomposition (MOEA/D) is widely used to dispose multiobjective optimization problems (MOPs). The performance of MOEA/D is not ideal when deals with MOPs which are with degenerated curve. Aiming to this weakness, an adaptive weight adjustment strategy named BPO is proposed in this paper. In order to solve this problem, BPO strategy is based on the number of reference points referencing the archive. The results of tests show that MOEA/D with BPO strategy can obtain ideal performance. MOEA/D with BPO strategy can improves the weakness of MOEA/D effectively on MOPs which are with degenerated curve.

Diagnosis and treatment of sub-health have the characteristics of multi-factors, uncertainty of pathological criteria and complexity of management methods. Therefore, the traditional rule-based expert reasoning method has limitations in the implementation of sub-health diagnosis scheme. Considering the characteristics of sub-health, this paper uses unstructured data elements to represent sub-health diagnosis case, and establishes an intelligent generation system CBR- SDC for sub-health diagnosis cases. The improved hybrid similarity retrieval method based on fuzzy mathematics and Euler distance, and the weight adjustment strategy and algorithm of PULL&PUSH are applied to case reasoning to solve the retrieval and correction of sub-health cases. The effectiveness of the method and the efficiency of case retrieval and revision are verified by comparative experiments. It provides an efficient new method for assistant diagnosis of sub-health pathology in medical industry.

This contribution presents an evaluation of the effectiveness of the Facebook social network promoting the brands of a number of selected Slovak wellness hotels based on the DEA (Data Envelopment Analysis) methodology and its selected models. The research question is that hotel guests use the funpages of the Slovak wellness hotels on the Facebook social network to learn more about its services and also how the Slovak wellness hotels use their funpages to promote their brand and communicate with their consumers. During the four months in 2018 (September – December), data on input and output variables was collected, with data from photos, videos and links to "funpage" hotels on Facebook and output to "Likes" and "Comments". The measurement of the efficiency of these input and output variables in order to assess the effectiveness of 16 wellness hotel brands operating in the Slovak Republic was based on an input-oriented CCR DEA model with weight adjustment via the Assurance Region. The number of Likes and comments on the Facebook pages of the 16 Slovak wellness hotels suggests that hotel guests use Facebook to learn more about the services and events they provide. The DEA model is therefore an effective tool to help evaluate the effectiveness of a business in a hotel sector on a social network, such as Facebook, in promoting its brands, as it uses multiple variables and does not necessarily require an input-output relationship. The results of using this method confirmed the research question.

The krill herd (KH) algorithm is an emerging meta-heuristic algorithm for solving complex problems. Though it is robust in optimization, there are some parameters that need to be fine-tuned for improved performance. This paper proposes an entropy-based inertia weight krill herd (EBIWKH) algorithm, which could adaptively adjust the inertia weight according to the variance of population entropy. It is tested on CEC2017 benchmark functions and compared with other inertia weight adjustment strategies. Experimental results show that the EBIWKH algorithm is more robust and stable than other compared algorithms.

Ensemble learning combines a series of base classifiers and the final result is assigned to the corresponding class by using a majority voting mechanism. However, the base classifiers in the ensemble cannot perform equally well. Therefore, the base classifiers should be assigned different weights in order to increase classification performance. In this study, a novel Weighted Majority Voting Ensemble (WMVE) approach is proposed, which evaluates individual performances of each classifier in the ensemble and adjusts their contributions to class decision. In the proposed weight adjustment model, only reward mechanism is provided, so punishment is not included. Classifiers that correctly classify observations which are not correctly classified by most of the classifiers gain more weights in the ensemble. In the experimental studies, increasing value of weight was calculated for each classifier in a heterogeneous collection of classification algorithms, including C4.5 decision tree, support vector machine, k-nearest neighbor, k-star, and naive Bayes. The proposed method (WMVE) was compared with the Simple Majority Voting Ensemble (SMVE) approach in terms of classification accuracy on 28 benchmark datasets. The effectiveness of the proposed method is confirmed by the experimental results.

In the automotive industry, the crashworthiness design of vehicles is of special importance. In this work, a multi-objective model for the vehicle design which minimizes three objectives, weight, acceleration characteristics, and toe-board intrusion, is considered, and a novel evolutionary algorithm based on decomposition and adaptive weight adjustment is designed to solve this problem. The experimental results reveal that the proposed algorithm works better than MOEA/D MOEA/D-AWA and NSGAII on this problem.

As the scale of the power system continues to expand, the power equipment fault rate gradually increases, which puts forward higher requirements for transformer fault diagnosis. Through fault diagnosis technology, faults can be found in advance during transformer operation, measures can be taken in time to reduce the possibility of accidents. It is found that the composition and content of dissolved gas in oil are closely related to the types of defects, and the composition and content of dissolved gas in transformer oil can play an important role in the prediction of transformer operating state and fault diagnosis. In order to obtain higher prediction accuracy, not only the composition and content of gas in oil, but also the correlation between fault types and characteristic quantities should be considered. In this paper, BP neural network algorithm based on association rules is adopted. Apriori algorithm of association rules reveals the association rules by mining high frequency terms in the data set of characteristic quantities. Apriori algorithm can effectively explore the confidence degree of association rules between feature quantity and running state and apply it as a weight value to the prediction link. Through the training steps of repeated forward training, reverse transfer and weight adjustment, the transformer operation state is finally determined.

The lateral body weight-shifting task is commonly used in therapeutic programs for patients with orthopedic complaints or hemiparesis. Although the patients usually support themselves using the upper limbs during the task, it is unclear whether the use of upper limbs affects performance accuracy of lateral body weight shifting. Therefore, the aim of this study was to investigate the effects of support by the upper limbs on performance accuracy, particularly on the central tendency effects (i.e. overshooting for light targets and undershooting for heavy targets). Twenty-three able-bodied, neurologically intact individuals, who were right-handed and right-footed participants performed the lateral body weight-shifting task to shift one-third or two-thirds of their body weight toward the left and right lower limbs using support by the upper limbs. The result of correlation coefficients between interlimb weight adjustment and errors showed that the use of the upper limbs and interlimb weight adjustment related to the enhanced central tendency effects. The use of upper limbs generally contributes toward stabilizing posture, however, this is not the case with performance of the lateral body weight shifting. Moreover, the effects of using the upper limb on performance accuracy differed among leftward and rightward weight shifting. This result might be owing to the dominant side of the hand/foot and hemisphere lateralization.

This paper considers the conceptual similarities and differences in weight adjustment steps known as nonresponse adjustment, post-stratification, and calibration. The distinction is based on the information requirements, i.e., whether the data necessary for the specific type of weight adjustment exist at the population level, at the level of the original sample that includes both respondents and nonrespondents, and/or at the respondent level. An illustrative example is provided where the different weights are constructed.

Figure 16 in the original publication was incomplete. The original article has been corrected.

This paper develops a comprehensive algorithm for multi-expert multi-criteria decision making problems considering quantitative and qualitative criteria in forms of benefit, cost or target types. We focus on using probabilistic linguistic term sets to express the qualitative evaluations due to their excellence in expressing complex individual and collective linguistic assessments. Firstly, we develop a target-based linear normalization technique and a target-based vector normalization technique. A weight adjustment method is proposed to achieve the tradeoff between criteria after normalization. Given that the two target-based normalization techniques have different advantages, we then propose a ranking method, which consists three subordinate models, based on these two target-based normalization approaches and three aggregation techniques. Reliable results of a multi-expert multi-criteria decision making problem are determined by integrating the subordinate utility values and the ranks of alternatives. The proposed method is implemented to solve the green enterprise ranking problems and the excavation scheme selection problem for shallow buried tunnels, respectively. The advantages of the proposed method are emphasized through comparative analyses with other ranking methods.

With the increase of the hidden layer, the weight update of the LSTM neural network model depends heavily on the gradient descent algorithm, and the convergence speed is slow, resulting in the local extremum of the weight adjustment, which affects the prediction performance of the model. Based on this, this paper proposes an optimized LSTM neural network model based on adaptive genetic algorithm (AGA-LSTM). In this model, the mean squared error is designed as the fitness function, and the adaptive genetic algorithm (AGA) is used to globally optimize the weights between neuron nodes of the LSTM model to improve the generalization ability. The experimental results show that, on the UCI dataset, the prediction accuracy of the AGA-LSTM model is greatly improved compared to the standard LSTM model, which verifies the rationality of the model.

In this paper, we propose a new approach for dense depth estimation based on multimodal stereo images. Our approach employs a combined cost function utilizing robust metrics and a transformation to an illumination independent representation. Additionally, we present a confidence based weighting scheme which allows a pixel-wise weight adjustment within the cost function. We demonstrate the capabilities of our approach using RGB- and thermal images. The resulting depth maps are evaluated by comparing them to depth measurements of a Velodyne HDL-64E LiDAR sensor. We show that our method outperforms current state of the art dense matching methods regarding depth estimation based on multimodal input images.

Credibility information gives an indication of which users are most likely to share relevant images on social network feeds and consequently may help estimating the relevance of an image for retrieval purposes. Considering multiple credibilty evidences has be shown as an effective method for image ranking. In order to select and combine multiple credibility descriptors, this work proposes a genetic algorithm-based automatic context-adaptive weight adjustment model. The experimental results show promising effectiveness when compared to the baseline.

Introduction Patients with Parkinson's disease (PD) are known to have decision-making impairments in tasks involving probabilistic information. How PD patients utilize task-relevant probabilistic advance information to plan and initiate common motor tasks like grasping has not yet been studied. Methods PD patients (n = 15, OFF medication) and control participants repeatedly grasped and lifted an object, the weight of which could be light, medium, or heavy. Visual cues provided explicit probabilistic information about the upcoming weight at the start of each grip-lift trial. This information allows the force of the grasping fingers to be scaled predictively so that it matches the likely weight, with a suitable rate of initial force increase. Deterministic cues announced the upcoming weight with certainty in other grip-lift trials. In a weight adjustment experiment, participants associated each probabilistic cue with a specific heaviness. Results The weight adjustment experiments showed that the probabilistic cues were understood correctly. However, PD patients utilized the probabilistic information significantly less than controls during the grip-lift task. Specifically, patients did not initiate their grasp more forcefully when probabilistic cues announced a high likelihood (66.7% probability) of a heavy weight, in contrast to controls. Thus, probabilistic cues that encouraged a more vigorous action had no effect in PD. Nevertheless, patients and controls scaled their forces appropriately when deterministic cues announced the forthcoming weights unambiguously. Conclusions PD patients do not invest a high movement effort to initiate a grip-lift unless the necessity of such a vigorous action initiation is decidedly clear.

When we buy items in online stores, it is common to face recommended items that meet our interest. These recommendation system help users not only to find out related items, but also find new things that may interest users. Recommendation system has been widely studied and various models has been suggested such as, collaborative filtering and content-based filtering. Though collaborative filtering shows good performance for predicting users preference, there are some conditions where collaborative filtering cannot be applied. Sparsity in user data causes problems in comparing users. Systems which are newly starting or companies having small number of users are also hard to apply collaborative filtering. Content-based filtering should be used to support this conditions, but content-based filtering has some drawbacks and weakness which are tendency of recommending similar items, and keeping history of a user makes recommendation simple and not able to follow up users preference changes. To overcome this drawbacks and limitations, we suggest weighted window assisted user history based recommendation system, which captures user’s purchase patterns and applies them to window weight adjustment. The system is capable of following current preference of a user, removing useless recommendation and suggesting items which cannot be simply found by users. To examine the performance under user and data sparsity environment, we applied data from start-up trading company. Through the experiments, we evaluate the operation of the proposed recommendation system.

With BDS receiver widely used in many fields, it faced some problems such as wide range of dynamic stress and thermal noise interference which caused by receiver racing and thermal shock of electrons in the process of tracking. Conventional receiver increase the bandwidth to expand the dynamic stress range, but it caused more thermal noise getting into the loop so that tracking accuracy decreased. In order to improve receiver tracking performance, a design scheme of the BDS receiver tracking loop based on the weight adjustment and adaptive bandwidth is proposed. This scheme utilizes line-of-sight dynamic of BDS receiver and carrier to noise power density ratio to adaptively adjust the bandwidth and the weight of the loop order so that the loop has a better noise inhibitory effect under the premise of stable tracking. The simulation results show that compared to the conventional receiver which is using fixed bandwidth and tracking loop order, this scheme can enhance the dynamic tracking performance and tracking accuracy.

Predicting and diagnosing various diseases are two effective components in medical science. Recently, with increasing expansion of science, using the decision support system can help significantly to the therapeutic policies of physician. For this purpose, by using artificial intelligent systems in this study, we are trying to predict and diagnose the breast cancer which is one of the most prevalent cancers among the women. In this study, three datasets of Wisconsin cancer diagnosis were used which were obtained by FNA test. Process of breast cancer diagnosis was investigated by a new approach. With the help of fuzzy rule based classification systems, datasets were classified. Rule weight learning algorithm has been used to improve the classification system which minimizing the classification error on the training data using the rule weight adjustment. An integrated method was also presented to reduce the dimensions of datasets features using the firefly algorithm and statistical repair mechanism. Comparing the performance of the aforementioned method with other ones shows the desired performance and high accuracy of the proposed method.

In this paper, an ensemble form of the semi-supervised Fisher Discriminant Analysis (FDA) model is developed for fault classification in industrial processes. This method uses the K Nearest Neighbor (KNN) algorithm to merge the metric level outputs, which are obtained by the sub-classifiers in the ensemble model, to get the final classification result. An adaptive form is further proposed to improve the classification performance by putting forward to a new method of weight adjustment. While semi-supervised learning can generate a better model by exploiting additional information contained in unlabeled data, ensemble learning achieves the promotion of algorithm robustness by integrating a series of weak learners. In addition, the property of diversity in ensemble learning can be boosted by incorporating different unlabeled data to different weak learners. Therefore, the combination of those two methods can achieve great generalization for the fault classification model. The performances of two proposed methods are evaluated through an industrial benchmark process.

The manual expert scheduling, currently utilized in shortwave position system, is unable to satisfy the modern scenario of intensive tasks and deficient resources, thus resulting in the low position accuracy. To address the problem, we propose a shortwave-position-resource scheduling method based on Back Propagating Neural Network (BPNN) and Dynamic Weight Adjustment (DWA). Firstly, we conceive a basic centralized optimization model for problem formulation, constructing an objective function consisted by Geometrical Dilution of Precision (GDOP) and signal quality via linear weight summation. Furthermore, we study the dynamic adjustment strategy of weights. With weights updated according to the gradients of factors, the objective function inclines to the factor of better performance during iterations, thus leading to a better optimization searching ability. Moreover, by introducing the Bayesian Regulation (BR) and Levenberg-Marquardt (LM) algorithm, we make an improvement on the generalization performance of BPNN, then achieving the signal quality estimation in the shortwave channel. Simulations indicate that the proposed method obtains an excellent robustness in scenes of different tasks number. When compared with other scheduling methods, the proposed method can effectively enhance the location accuracy. Experiments also show that the timeliness of the method is relatively low and needs further improvement.

Background Patients (pts) with RA may be able to switch from IV to SC abatacept with no loss of efficacy or safety concerns, but data are inconclusive. 1–4 In the ACTION (AbataCepTIn rOutiNe clinical practice; NCT02109666) study, a 1-year interim analysis showed that switching had no adverse clinical effect. 5 Objectives To examine treatment patterns and explore abatacept formulation switching over 2 years in ACTION. Methods ACTION is a 2-year, prospective, observational study of pts with moderate-to-severe RA who initiated IV abatacept in Europe and Canada between May 2008 and December 2013. Assessments in biologic-naive and biologic-failure pts were: baseline characteristics, rates of and reasons for switching (IV to SC), and re-switching to IV over 2 years. Descriptive data were generated: mean (SD) for continuous variables and n (%) for categorical variables. Rates of switching were estimated by Kaplan–Meier analysis. Cohorts were pooled to analyse further pts who switched owing to low numbers. Results In the ACTION cohort, 2350/2364 pts (99.4%) were evaluable for this analysis (673 [28.6%] biologic naive, 1677 [71.4%] biologic failure). A total of 729 (43.4%) biologic-failure pts had received 1, and 948 (56.6%) had received ≥2 previous biologics. Baseline characteristics in biologic-naive and biologic-failure pts, respectively, were: mean (SD) age 59.9 (12.7) and 56.9 (12.5) years; RA duration 7.2 (8.22) and 12.1 (9.13) years; 496 (73.7%) and 1379 (82.2%) were women; 621 (92.3%) and 1552 (92.5%) had received prior MTX; and 533 (79.2%) and 1386 (82.6%) had received corticosteroids. Over 2 years, 195 pts switched from IV to SC abatacept (57 biologic naive, 138 biologic failure; Fig.). Reasons for switching were available for 172 pts (51 biologic naive, 121 biologic failure; some had >1 reason); biologic naive/biologic failure: pt wish 54.9%/62.0%, physician choice 31.4%/19.8%, safety 5.9%/9.9%, remission/major improvement 3.9%/5.0%, poor compliance 0%/4.1%, lack of efficacy 2.0%/3.3%, surgery 2.0%/0.8%, weight adjustment 2.0%/0%, other 49.0%/36.4%. Only eight pts (2.6%) re-switched to IV abatacept (2 biologic naive, 6 biologic failure). Reasons for re-switching were: pt wish (n=4), lack of efficacy (n=4), safety issue (n=1) and other (n=2). Conclusions Less than 5% of pts who switched formulation from IV to SC abatacept in real-world clinical practice re-switched to the IV formulation, suggesting that switching has no adverse clinical impact. A change in formulation was mainly due to pt wish, reflecting their involvement in decision-making. References Keystone EC, et al. Ann Rheum Dis 2012;71:857–61. Mueller R, et al. Arthritis Rheumatol 2015;67(Suppl. 10):651–52. Reggia R, et al. J Rheumatol 2015;42:193–5. Monti S, et al. J Rheumatol 2015;42:1993–4. Alten R, et al. Ann Rheum Dis 2016;75(Suppl. 2):202. Disclosure of Interest R. Alten Grant/research support from: Bristol-Myers Squibb, Speakers bureau: Bristol-Myers Squibb, H.-M. Lorenz Consultant for: AbbVie, Bristol-Myers Squibb, Roche-Chugai, UCB, MSD, GSK, SOBI, Medac, Novartis, Janssen-Cilag, Astra-Zeneca, Pfizer, Actelion, X. Mariette Grant/research support from: Biogen, Pfizer, UCB, Consultant for: Bristol-Myers Squibb, LFB, Pfizer, GSK, UCB, H. Nuslein Consultant for: AbbVie, Bristol-Myers Squibb, Celgene, Janssen, Lilly, MSD, Novartis, Pfizer, Roche, Speakers bureau: AbbVie, Bristol-Myers Squibb, Celgene, Janssen, Lilly, MSD, Novartis, Pfizer, Roche, M. Galeazzi: None declared, F. Navarro Grant/research support from: Pfizer, MSD, AbbVie, Bristol-Myers Squibb, Roche, Consultant for: Pfizer, MSD, Roche, UCB, AbbVie, Bristol-Myers Squibb, Jansen, Lilly, Speakers bureau: Pfizer, MSD, Roche, UCB, AbbVie, Bristol-Myers Squibb, M. Chartier Employee of: Bristol-Myers Squibb, J. Heitzmann: None declared, C. Rauch Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb, M. Le Bars Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb

Use of Murthy’s method in estimation of population parameters, such as population totals, population means, and population variances has been limited to surveys where survey data values are complete. This study applies weight adjustment technique to estimate a population total under simple random sampling without replacement. The asymptotic properties show that the estimated population total is sufficient for the true population total. The proposed estimator is obtained by symmetrizing Murthy’s estimator.

Objective The aim of this study was to determine the impact on the risk calculation of various ways of handling maternal weight when these data are provided in the first part but not the second part of a sequential screening protocol. Method A retrospective analysis of 38 986 sequential screens was carried out in which weight was provided in both the first and second trimesters. Three potential strategies for calculating multiples of the median values when the weight is not recorded at the time of second trimester risk evaluation were evaluated. First, perform no weight adjustment. Second, use the first trimester weight. Third, use the predicted second trimester weight on the basis of the first trimester weight. To predict the second trimester weight, we used a random-effects, multi-level model. Results The screen positive rate for Down syndrome was 3.0% (1151/38 986) and trisomy 18 alone 0.12% (47/38 986). The three strategies resulted in 196 (0.50%), 41 (0.11%), and 23 (0.06%) patients switching risk categories with the no adjustment, first trimester weight, and predicted weight strategies, respectively. Conclusion Utilizing the first trimester weight or the predicted second trimester weight in sequential screening when second trimester weight is not provided offers an affordable alternative for laboratories to provide robust risk calculations and interpretations without requiring excessive use of resources. © 2014 John Wiley & Sons, Ltd.

This paper studies control strategies for adaptive cruise control (ACC) systems in battery electric vehicles (BEVs). A hierarchical control structure is adopted for the ACC system, and the structure contains an upper controller and a lower controller. This paper focuses on the upper controller. In the upper controller, model predictive control (MPC) is applied for optimizing multiple objectives in the car-following process. In addition, multiple objectives, including safety, tracking, comfort, and energy economy, can be transformed into a symmetric objective function with constraints in MPC. In the objective function, the corresponding weight matrix for the optimization of multiple objectives is implemented in symmetric form to reduce the computational complexity. The weights in the weight matrix are usually set to be constant. However, the motion states of the own vehicle and the front vehicle change with respect to time during a car-following process, resulting in variation of the driving conditions. MPCs with constant weights do not adapt well to changes in driving conditions, which limits the performance of the ACC system. Therefore, a strategy for weight adjustment is proposed in order to improve the tracking performance, in which some weights in MPC can be adjusted according to the relative velocity of two vehicles in real time. The simulation experiments are carried out to demonstrate the effectiveness of the strategy for weight adjustment. Based on achieving the other control objectives, the ACC system with the weight adjustment has better tracking performance than the ACC system with the constant weight. While the tracking is improved, the energy economy is also improved.

In neural networks, the accuracies of its networks are mainly relying on two important factors which are the centers and the networks weight. The gradient descent algorithm is a widely used weight adjustment algorithm in most of neural networks training algorithm. However, the method is known for its weakness for easily trap in local minima. It suffers from a random weight generated for the networks during initial stage of training at input layer to hidden layer networks. The performance of radial basis function networks (RBFN) has been improved from different perspectives, including centroid initialization problem to weight correction stage over the years. Unfortunately, the solution does not provide a good trade-off between quality and efficiency of the weight produces by the algorithm. To solve this problem, an improved gradient descent algorithm for finding initial weight and improve the overall networks weight is proposed. This improved version algorithm is incorporated into RBFN training algorithm for updating weight. Hence, this paper presented an improved RBFN in term of algorithm for improving the weight adjustment in RBFN during training process. The proposed training algorithm, which uses improved gradient descent algorithm for weight adjustment for training RBFN, obtained significant improvement in predictions compared to the standard RBFN. The proposed training algorithm was implemented in MATLAB environment. The proposed improved network called IRBFN was tested against the standard RBFN in predictions. The experimental models were tested on four literatures nonlinear function and four real-world application problems, particularly in Air pollutant problem, Biochemical Oxygen Demand (BOD) problem, Phytoplankton problem, and forex pair EURUSD. The results are compared to IRBFN for root mean square error (RMSE) values with standard RBFN. The IRBFN yielded a promising result with an average improvement percentage more than 40 percent in RMSE.

Evolving fuzzy rule-based systems are very powerful methods for online classification of data streams. In these systems, the classifier is updated by generating, removing and modifying fuzzy classification rules. One of the simplest but efficient algorithms of this type is evolving classifier (eClass) that generates fuzzy classification rules without any prior knowledge. However, this algorithm cannot cope properly with drift and shift in the concept of data. In order to improve the performance of this algorithm, in this paper, we propose a scheme that assigns a weight to each fuzzy rule and propose a new efficient online method to adjust the weight of fuzzy classification rules. Using the proposed rule-weighting algorithm, the classifier can quickly cope with drift and shift in the concept of data. Our algorithm is in fact the modified version of a batch mode rule-weight learning algorithm proposed in the past to be consistent with characteristics of data streams. We use some real life and some synthetic datasets to assess the performance of our algorithm in comparison with eClass and some other methods proposed in the past for handling data streams. The results of experiments show that our proposed algorithm performs significantly better than eClass and other methods proposed in the past for classification of data streams.

The traditional search engine is unable to correct search for the magnanimous information in Deep Web hides. The Web database's classification is the key step which integrates with the Web database classification and retrieves. This article has proposed one kind of classification based on machine learning's web database. The experiment has indicated that after this taxonomic approach undergoes few sample training, it can achieve the very good classified effect, and along with training sample's increase, this classifier's performance maintains stable and the rate of accuracy and the recalling rate fluctuate in the very small scope.

Imputation is a commonly used method to handle missing survey data. The performance of the imputation method is influenced by various factors, especially an outlier. The removal of the outlier in a data set is a simple and effective approach to reduce the effect of an outlier. In this paper in order to improve the precision of multiple imputation, we study a imputation method which reduces the effect of outlier using various weight adjustment methods that include the removal of an outlier method. The regression method in PROC/MI in SAS is used for multiple imputation and the obtained final adjusted weight is used as a weight variable to obtain the imputed values. Simulation studies compared the performance of various weight adjustment methods and Monthly Labor Statistic data is used for real data analysis.