A New Speed Limit Recognition Methodology Based on Ensemble Learning:Hardware Validation

Advanced DriverAssistance Systems(ADAS)technologies can assist drivers or be part of automatic driving systems to support the driving process and improve the level of safety and comfort on the road.Traffic Sign Recognition System(TSRS)is one of themost important components ofADAS.Among the challengeswith TSRS is being able to recognize road signs with the highest accuracy and the shortest processing time.Accordingly,this paper introduces a new real time methodology recognizing Speed Limit Signs based on a trio of developed modules.Firstly,the Speed Limit Detection(SLD)module uses the Haar Cascade technique to generate a new SL detector in order to localize SL signs within captured frames.Secondly,the Speed Limit Classification(SLC)module,featuring machine learning classifiers alongside a newly developed model called DeepSL,harnesses the power of a CNN architecture to extract intricate features from speed limit sign images,ensuring efficient and precise recognition.In addition,a new Speed Limit Classifiers Fusion(SLCF)module has been developed by combining trained ML classifiers and the DeepSL model by using the Dempster-Shafer theory of belief functions and ensemble learning’s voting technique.Through rigorous software and hardware validation processes,the proposedmethodology has achieved highly significant F1 scores of 99.98%and 99.96%for DS theory and the votingmethod,respectively.Furthermore,a prototype encompassing all components demonstrates outstanding reliability and efficacy,with processing times of 150 ms for the Raspberry Pi board and 81.5 ms for the Nano Jetson board,marking a significant advancement in TSRS technology.

A redundant subspace weighting procedure for clock ensemble

A redundant-subspace-weighting(RSW)-based approach is proposed to enhance the frequency stability on a time scale of a clock ensemble.In this method,multiple overlapping subspaces are constructed in the clock ensemble,and the weight of each clock in this ensemble is defined by using the spatial covariance matrix.The superimposition average of covariances in different subspaces reduces the correlations between clocks in the same laboratory to some extent.After optimizing the parameters of this weighting procedure,the frequency stabilities of virtual clock ensembles are significantly improved in most cases.

Ensemble Bayesian method for parameter distribution inference:application to reactor physics

The estimation of model parameters is an important subject in engineering.In this area of work,the prevailing approach is to estimate or calculate these as deterministic parameters.In this study,we consider the model parameters from the perspective of random variables and describe the general form of the parameter distribution inference problem.Under this framework,we propose an ensemble Bayesian method by introducing Bayesian inference and the Markov chain Monte Carlo(MCMC)method.Experiments on a finite cylindrical reactor and a 2D IAEA benchmark problem show that the proposed method converges quickly and can estimate parameters effectively,even for several correlated parameters simultaneously.Our experiments include cases of engineering software calls,demonstrating that the method can be applied to engineering,such as nuclear reactor engineering.

Statistical Downscaling for Multi-Model Ensemble Prediction of Summer Monsoon Rainfall in the Asia-Pacific Region Using Geopotential Height Field

The 21-yr ensemble predictions of model precipitation and circulation in the East Asian and western North Pacific （Asia-Pacific） summer monsoon region （0°-50°N, 100° 150°E） were evaluated in nine different AGCM, used in the Asia-Pacific Economic Cooperation Climate Center （APCC） multi-model ensemble seasonal prediction system. The analysis indicates that the precipitation anomaly patterns of model ensemble predictions are substantially different from the observed counterparts in this region, but the summer monsoon circulations are reasonably predicted. For example, all models can well produce the interannual variability of the western North Pacific monsoon index （WNPMI） defined by 850 hPa winds, but they failed to predict the relationship between WNPMI and precipitation anomalies. The interannual variability of the 500 hPa geopotential height （GPH） can be well predicted by the models in contrast to precipitation anomalies. On the basis of such model performances and the relationship between the interannual variations of 500 hPa GPH and precipitation anomalies, we developed a statistical scheme used to downscale the summer monsoon precipitation anomaly on the basis of EOF and singular value decomposition （SVD）. In this scheme, the three leading EOF modes of 500 hPa GPH anomaly fields predicted by the models are firstly corrected by the linear regression between the principal components in each model and observation, respectively. Then, the corrected model GPH is chosen as the predictor to downscale the precipitation anomaly field, which is assembled by the forecasted expansion coefficients of model 500 hPa GPH and the three leading SVD modes of observed precipitation anomaly corresponding to the prediction of model 500 hPa GPH during a 19-year training period. The cross-validated forecasts suggest that this downscaling scheme may have a potential to improve the forecast skill of the precipitation anomaly in the South China Sea, western North Pacific and the East Asia Pacific regions, where the anomaly correlation coefficient （ACC） has been improved by 0.14, corresponding to the reduced RMSE of 10.4% in the conventional multi-model ensemble （MME） forecast.

STUDY OF THE MODIFICATION OF MULTI-MODEL ENSEMBLE SCHEMES FOR TROPICAL CYCLONE FORECASTS

This study investigates multi-model ensemble forecasts of track and intensity of tropical cyclones over the western Pacific, based on forecast outputs from the China Meteorological Administration, European Centre for Medium-Range Weather Forecasts, Japan Meteorological Agency and National Centers for Environmental Prediction in the THORPEX Interactive Grand Global Ensemble(TIGGE) datasets. The multi-model ensemble schemes, namely the bias-removed ensemble mean(BREM) and superensemble(SUP), are compared with the ensemble mean(EMN) and single-model forecasts. Moreover, a new model bias estimation scheme is investigated and applied to the BREM and SUP schemes. The results showed that, compared with single-model forecasts and EMN, the multi-model ensembles of the BREM and SUP schemes can have smaller errors in most cases. However, there were also circumstances where BREM was less skillful than EMN, indicating that using a time-averaged error as model bias is not optimal. A new model bias estimation scheme of the biweight mean is introduced. Through minimizing the negative influence of singular errors, this scheme can obtain a more accurate model bias estimation and improve the BREM forecast skill. The application of the biweight mean in the bias calculation of SUP also resulted in improved skill. The results indicate that the modification of multi-model ensemble schemes through this bias estimation method is feasible.

Ensemble Simulation of Land Evapotranspiration in China Based on a Multi-Forcing and Multi-Model Approach

In order to reduce the uncertainty of offline land surface model （LSM） simulations of land evapotranspiration （ET）, we used ensemble simulations based on three meteorological forcing datasets [Princeton, ITPCAS （Institute of Tibetan Plateau Research, Chinese Academy of Sciences）, Qian] and four LSMs （BATS, VIC, CLM3.0 and CLM3.5）, to explore the trends and spatiotemporal characteristics of ET, as well as the spatiotemporal pattern of ET in response to climate factors over China's Mainland during 1982-2007. The results showed that various simulations of each member and their arithmetic mean （EnsAVlean） could capture the spatial distribution and seasonal pattern of ET sufficiently well, where they exhibited more significant spatial and seasonal variation in the ET compared with observation-based ET estimates （Obs_MTE）. For the mean annual ET, we found that the BATS forced by Princeton forcing overestimated the annual mean ET compared with Obs_MTE for most of the basins in China, whereas the VIC forced by Princeton forcing showed underestimations. By contrast, the Ens_Mean was closer to Obs_MTE, although the results were underestimated over Southeast China. Furthermore, both the Obs_MTE and Ens_Mean exhibited a significant increasing trend during 1982-98; whereas after 1998, when the last big EI Nifio event occurred, the Ens_Mean tended to decrease significantly between 1999 and 2007, although the change was not significant for Obs_MTE. Changes in air temperature and shortwave radiation played key roles in the long-term variation in ET over the humid area of China, but precipitation mainly controlled the long-term variation in ET in arid and semi-arid areas of China.

Improving Multi-model Ensemble Probabilistic Prediction of Yangtze River Valley Summer Rainfall

Seasonal prediction of summer rainfall over the Yangtze River valley（YRV） is valuable for agricultural and industrial production and freshwater resource management in China, but remains a major challenge. Earlier multi-model ensemble（MME） prediction schemes for summer rainfall over China focus on single-value prediction, which cannot provide the necessary uncertainty information, while commonly-used ensemble schemes for probability density function（PDF） prediction are not adapted to YRV summer rainfall prediction. In the present study, an MME PDF prediction scheme is proposed based on the ENSEMBLES hindcasts. It is similar to the earlier Bayesian ensemble prediction scheme, but with optimization of ensemble members and a revision of the variance modeling of the likelihood function. The optimized ensemble members are regressed YRV summer rainfall with factors selected from model outputs of synchronous 500-h Pa geopotential height as predictors. The revised variance modeling of the likelihood function is a simple linear regression with ensemble spread as the predictor. The cross-validation skill of 1960–2002 YRV summer rainfall prediction shows that the new scheme produces a skillful PDF prediction, and is much better-calibrated, sharper, and more accurate than the earlier Bayesian ensemble and raw ensemble.

A COMBINED VERIFICATION METHOD FOR PREDICTABILITY OF PERSISTENT HEAVY RAINFALL EVENTS OVER EAST ASIA BASED ON ENSEMBLE FORECAST

Persistent Heavy Rainfall(PHR)is the most influential extreme weather event in Asia in summer,and thus it has attracted intensive interests of many scientists.In this study,operational global ensemble forecasts from China Meteorological Administration(CMA)are used,and a new verification method applied to evaluate the predictability of PHR is investigated.A metrics called Index of Composite Predictability(ICP)established on basic verification indicators,i.e.,Equitable Threat Score(ETS)of 24 h accumulated precipitation and Root Mean Square Error(RMSE)of Height at 500 h Pa,are selected in this study to distinguish"good"and"poor"prediction from all ensemble members.With the use of the metrics of ICP,the predictability of two typical PHR events in June 2010 and June 2011 is estimated.The results show that the"good member"and"poor member"can be identified by ICP and there is an obvious discrepancy in their ability to predict the key weather system that affects PHR."Good member"shows a higher predictability both in synoptic scale and mesoscale weather system in their location,duration and the movement.The growth errors for"poor"members is mainly due to errors of initial conditions in northern polar region.The growth of perturbation errors and the reason for better or worse performance of ensemble member also have great value for future model improvement and further research.

An Approach to Extract Effective Information of Monthly Dynamical Prediction-The Use of Ensemble Method

The approach of getting useful information of monthly dynamical prediction from ensemble forecasts is studied. The extended range ensemble forecasts (8 members, the initial perturbations of the lagged average forecast (LAF)(0000, 0600, 1200 and 1800 GMT in two consecutive days) of the 500 hPa height field with the global spectral model (T63L16) from January to May 1997 are provided by the National Climate Center of China. The relationship between the spread of ensemble measured by root–mean–square deviation of ensemble member from ensemble mean and forecast skill (the anomaly correlation or the root–mean–square distance between the ensemble mean forecast and the observation) is significant. The spread of ensemble can evaluate the useful forecast days N for the best estimate of 30 days mean. Thus, a weighted mean approach based on ensemble spread is put forward for monthly dynamical prediction. The anomaly correlation of the weighted monthly mean by the ensemble spread is higher than that of both the arithmetic mean and the linear weighted mean. Better results of the monthly mean circulation and anomaly are obtained from the ensemble spread weighted mean. Key words Monthly prediction - Ensemble method - Spread of ensemble Supported by the Excellent National State Key Laboratory Project (49823002), the National Key Project ‘Study on Chinese Short-Term Climate Forecast System’ (96-908-02) and IAP Innovation Foundation (8-1308).The data were provided through the National Climate Center of China. The authors wish to thank Ms. Chen Lijuan for her assistance.

A Hybrid Neural Network Model for Marine Dissolved Oxygen Concentrations Time-Series Forecasting Based on Multi-Factor Analysis and a Multi-Model Ensemble

Dissolved oxygen(DO)is an important indicator of aquaculture,and its accurate forecasting can effectively improve the quality of aquatic products.In this paper,a new DO hybrid forecasting model is proposed that includes three stages:multi-factor analysis,adaptive decomposition,and an optimizationbased ensemble.First,considering the complex factors affecting DO,the grey relational(GR)degree method is used to screen out the environmental factors most closely related to DO.The consideration of multiple factors makes model fusion more effective.Second,the series of DO,water temperature,salinity,and oxygen saturation are decomposed adaptively into sub-series by means of the empirical wavelet transform(EWT)method.Then,five benchmark models are utilized to forecast the sub-series of EWT decomposition.The ensemble weights of these five sub-forecasting models are calculated by particle swarm optimization and gravitational search algorithm(PSOGSA).Finally,a multi-factor ensemble model for DO is obtained by weighted allocation.The performance of the proposed model is verified by timeseries data collected by the pacific islands ocean observing system(PacIOOS)from the WQB04 station at Hilo.The evaluation indicators involved in the experiment include the Nash–Sutcliffe efficiency(NSE),Kling–Gupta efficiency(KGE),mean absolute percent error(MAPE),standard deviation of error(SDE),and coefficient of determination(R^(2)).Example analysis demonstrates that:①The proposed model can obtain excellent DO forecasting results;②the proposed model is superior to other comparison models;and③the forecasting model can be used to analyze the trend of DO and enable managers to make better management decisions.

Construction of Parsimonious Event Risk Scores by an Ensemble Method. An Illustration for Short-Term Predictions in Chronic Heart Failure Patients from the GISSI-HF Trial

Selecting which explanatory variables to include in a given score is a common difficulty, as a balance must be found between statistical fit and practical application. This article presents a methodology for constructing parsimonious event risk scores combining a stepwise selection of variables with ensemble scores obtained by aggregation of several scores, using several classifiers, bootstrap samples and various modalities of random selection of variables. Selection methods based on a probabilistic model can be used to achieve a stepwise selection for a given classifier such as logistic regression, but not directly for an ensemble classifier constructed by aggregation of several classifiers. Three selection methods are proposed in this framework, two involving a backward selection of the variables based on their coefficients in an ensemble score and the third involving a forward selection of the variables maximizing the AUC. The stepwise selection allows constructing a succession of scores, with the practitioner able to choose which score best fits his needs. These three methods are compared in an application to construct parsimonious short-term event risk scores in chronic HF patients, using as event the composite endpoint of death or hospitalization for worsening HF within 180 days of a visit. Focusing on the fastest method, four scores are constructed, yielding out-of-bag AUCs ranging from 0.81 (26 variables) to 0.76 (2 variables).

A Bayesian Scheme for Probabilistic Multi-Model Ensemble Prediction of Summer Rainfall over the Yangtze River Valley

A Bayesian probabilistic prediction scheme of the Yangtze River Valley (YRV) summer rainfall is proposed to combine forecast information from multi-model ensemble dataset provided by ENSEMBLES project.Due to the low forecast skill of rainfall in dynamic models,the time series of regressed YRV summer rainfall are selected as ensemble members in the new scheme,instead of commonly-used YRV summer rainfall simulated by models.Each time series of regressed YRV summer rainfall is derived from a simple linear regression.The predictor in each simple linear regression is the skillfully simulated circulation or surface temperature factor which is highly linear with the observed YRV summer rainfall in the training set.The high correlation between the ensemble mean of these regressed YRV summer rainfall and observation benefit extracting more sample information from the ensemble system.The results show that the cross-validated skill of the new scheme over the period of 1960 to 2002 is much higher than equally-weighted ensemble,multiple linear regression,and Bayesian ensemble with simulated YRV summer rainfall as ensemble members.In addition,the new scheme is also more skillful than reference forecasts (random forecast at a 0.01 significance level for ensemble mean and climatology forecast for probability density function).

An ensemble-based SST nudging method proposed for correcting the subsurface temperature field in climate model

An ensemble-based assimilation method is proposed for correcting the subsurface temperature field when nudging the sea surface temperature(SST) observations into the Max Planck Institute(MPI) climate model,ECHAM5/MPI-OM. This method can project SST directly to subsurface according to model ensemble-based correlations between SST and subsurface temperature. Results from a 50 year(1960–2009) assimilation experiment show the method can improve the subsurface temperature field up to 300 m compared to the qualitycontrolled subsurface ocean temperature objective analyses(EN4), through reducing the biases of the thermal states, improving the thermocline structure, and reducing the root mean square(RMS) errors. Moreover, as most of the improvements concentrate over the upper 100 m, the ocean heat content in the upper 100 m(OHT100 m)is further adopted as a property to validate the performance of the ensemble-based correction method. The results show that RMS errors of the global OHT100 m convergent to one value after several times iteration,indicating this method can represent the relationship between SST and subsurface temperature fields well, and then improve the accuracy of the simulation in the subsurface temperature of the climate model.

Validation of the effects of temperature simulated by multi-model ensemble and prediction of mean temperature changes for the next three decades in China

Using series of daily average temperature observations over the period of 1961-1999 of 701 meteorological stations in China, and simulated results of 20 global climate models （such as BCCR_BCM2.0, CGCM3T47） during the same period as the observation, we validate and analyze the simulated results of the models by using three factor statistical method, achieve the results of mul- ti-model ensemble, test and verify the results of multi-model ensemble by using the observation data during the period of 1991-1999. Finally, we analyze changes of the annual mean temperature result of multi-mode ensemble prediction for the period of 2011-2040 under the emission scenarios A2, A1B and B 1. Analyzed results show that： （1） Global climate models can repro- duce Chinese regional spatial distribution of annual mean temperature, especially in low latitudes and eastern China. （2） With the factor of the trend of annual mean temperature changes in reference period, there is an obvious bias between the model and the observation. （3） Testing the result of multi-model ensemble during the period of 1991-1999, we can simulate the trend of temper- ature increase. Compared to observation, the result of different weighing multi-model ensemble prediction is better than the same weighing ensemble. （4） For the period of 20ll-2040, the growth of the annual mean temperature in China, which results from multi-mode ensemble prediction, is above 1℃. In the spatial distribution of annual mean temperature, under the emission scenarios of A2, A1B and B 1, the trend of growth in South China region is the smallest, the increment is less than or equals to 0.8℃; the trends in the northwestern region and south of the Qinghai-Tibet Plateau are the largest, the increment is more than 1℃.

Study of perturbing method in regional BGM ensemble prediction system

Based on an Ensemble Prediction System with the BGM method on the regional numerical prediction model AREM, Static State Perturbation （SSP, including Initial Random Perturbation and Perturbation Restriction） and Dynamic State Perturbation （DSP） are designed. The impacts of both perturbations on precipitation prediction are studied by analyzing a slrong precipitation process oc- curring during July 20-21, 2008. The results show that both SSP and DSP play a positive role in prediction of mesoscale precipita- tion, such as lowering the （missing） rate of precipitation prediction. SSP is mainly helpful for the 24-hour prediction, while DSP can improve both 24-hour and 48-hour prediction. DSP is better than the two SSPs in the hit rate of regional precipitation prediction. However, the former also has a little higher false alarm rate than the latter. DSP enlarges in some extent the dispersion of EPS, which is good for EPS.

Using Hybrid and Diversity-Based Adaptive Ensemble Method for Binary Classification

This paper proposes an adaptive and diverse hybrid-based ensemble method to improve the performance of binary classification. The proposed method is a non-linear combination of base models and the application of adaptive selection of the most suitable model for each data instance. Ensemble method, an important machine learning technique uses multiple single models to construct a hybrid model. A hybrid model generally performs better compared to a single individual model. In a given dataset the application of diverse single models trained with different machine learning algorithms will have different capabilities in recognizing patterns in the given training sample. The proposed approach has been validated on Repeat Buyers Prediction dataset and Census Income Prediction dataset. The experiment results indicate up to 18.5% improvement on F1 score for the Repeat Buyers dataset compared to the best individual model. This improvement also indicates that the proposed ensemble method has an exceptional ability of dealing with imbalanced datasets. In addition, the proposed method outperforms two other commonly used ensemble methods (Averaging and Stacking) in terms of improved F1 score. Finally, our results produced a slightly higher AUC score of 0.718 compared to the previous result of AUC score of 0.712 in the Repeat Buyers competition. This roughly 1% increase AUC score in performance is significant considering a very big dataset such as Repeat Buyers.

EXPERIMENTS OF ENSEMBLE FORECAST OF TYPHOON TRACK USING BDA PERTURBING METHOD

A new method, BDA perturbing, is used in ensemble forecasting of typhoon track. This method is based on the Bogus Data Assimilation scheme. It perturbs the initial position and intensity of typhoons and gets; a series of bogus vortex. Then each bogus vortex is used in data assimilation to obtain initial conditions. Ensemble forecast members are constructed by conducting simulation with these initial conditions. Some cases of typhoon are chosen to test the validity of this new method and the results show that： using the BDA perturbing method to perturb initial position and intensity of typhoon for track tbrecast can improve accuracy, compared with the direct use of the BDA assimilation scheme. And it is concluded that a perturbing amplitude of intensity of 5 hPa is probably more appropriate than 10 hPa if the BDA perturbing method is used in combination with initial position perturbation.

A rough sets based pruning method for bagging ensemble

Ensemble techniques train a set of component classifiers and then combine their predictions to classify new patterns.Bagging is one of the most popular ensemble techniques for improving weak classifiers.However,it is hard to deploy in many real applications because of the large memory requirement and high computation cost to store and vote the predictions of component classifiers.Rough set theory is a formal mathematical tool to deal with incomplete or imprecise information,which has attracted a lot of attention from theory and application fields.In this paper,a novel rough sets based method is proposed to prune the classifiers obtained from bagging ensemble and select a subset of the component classifiers for aggregation.Experiment results show that the proposed method not only decreases the number of component classifiers but also obtains acceptable performance.

Coupling Ensemble Kalman Filter with Four-dimensional Variational Data Assimilation

This study examines the performance of coupling the deterministic four-dimensional variational assimilation system （4DVAR） with an ensemble Kalman filter （EnKF） to produce a superior hybrid approach for data assimilation. The coupled assimilation scheme （E4DVAR） benefits from using the state-dependent uncertainty provided by EnKF while taking advantage of 4DVAR in preventing filter divergence： the 4DVAR analysis produces posterior maximum likelihood solutions through minimization of a cost function about which the ensemble perturbations are transformed, and the resulting ensemble analysis can be propagated forward both for the next assimilation cycle and as a basis for ensemble forecasting. The feasibility and effectiveness of this coupled approach are demonstrated in an idealized model with simulated observations. It is found that the E4DVAR is capable of outperforming both 4DVAR and the EnKF under both perfect- and imperfect-model scenarios. The performance of the coupled scheme is also less sensitive to either the ensemble size or the assimilation window length than those for standard EnKF or 4DVAR implementations.

Strip flatness prediction of cold rolling based on ensemble methods

Aiming at the problem of insufficient prediction accuracy of strip flatness at the outlet of cold tandem rolling,the prediction performance of strip flatness based on different ensemble methods was studied and a high-precision prediction ensemble model of strip flatness at the outlet was established.Firstly,based on linear regression(LR),K nearest neighbors(KNN),support vector regression,regression trees(RT),and backpropagation neural network(BPN),bagging,boosting,and stacking ensemble methods were used for ensemble experiments.Secondly,three existing ensemble models,i.e.,random forest,extreme random tree(ET)and extreme gradient boosting,were used to conduct experiments and compare the results.The research shows that bagging,boosting,and stacking three ensemble methods have the most significant improvement in the prediction accuracy of the regression trees model,which is increased by 5.28%,6.51%,and 5.32%,respectively.At the same time,the stacking ensemble method improves both the simple model and the complex model,and the improvement effect on the simple base model is the greatest,which is 4.69%higher than that of the base model KNN.Comparing all of the ensemble models,the stacking ensemble model of level-1(ET,AdaBoost-RT,LR,BPN)paired with level-2(LR)was discovered to be the best model(EALB-LR)and can be further studied for industrial applications.