Hybrid Machine Learning Algorithms Research Articles

A new artificial multi-neural approach to estimate the hourly global solar radiation in a semi-arid climate site

In this study, we propose a new hybrid machine-learning algorithm called artificial multi-neural approach, for developing a multi-input single-output (MISO) model to estimate the hourly global solar radiation (HGSR) time series in Agdal site (latitude 31° 37′ N, longitude 08° 01′ W, elevation 466 m), Marrakesh, Morocco. To achieve this goal, three training algorithms (scaled conjugate gradient, Levenberg-Marquardt, and resilient backpropagation) are selected to train the developed model, using some accessible hourly meteorological variables recorded during 7 years (from 2008 to 2014) as exogenous inputs. Furthermore, a pertinence determination test (PDT) is used to point out the most pertinent inputs for the accurate HGSR estimation. The best configuration of the multi-neural model includes five known input parameters as the best scenario, i.e., acquisition hour, air temperature, relative humidity, wind speed, and precipitation. Some statistical indicators are used to evaluate the performance of the developed model. The obtained results demonstrate the reliability and the precision of our proposed machine-learning algorithm, representing a good solution for estimating solar radiation time series needed to design and manage solar energy systems compared to the present standards.

A Delay and Energy Efficient Multicast Routing Protocol using IWO and MOLO Algorithm for Vehicular Networks

Vehicular networks are significantly improving wireless communication network which provides an intelligent transportation system services among faster moving vehicles with internet and brings safety and comfort drive. From a single source vehicle the multicast routing protocols delivers multicast messages to all members of the multicast group by means of multi-hop communication. Increasing the density of vehicles results in channel overload, which increases the probability of data collision; hence reduction in successful received data will increase the delay. Therefore, delay and energy consumption are the major constraints that should affect the performance of routing. In this paper we suggest a delay and energy efficient multicast routing (DEMR) protocol for vehicular network using a hybrid machine learning algorithm. The DEMR protocol consists of four layers; are vehicle layer, fog layer, OpenFlow switch layer, and SDN controller layer. Moreover, to partition the vehicle layer, and select the optimal multicast path based on multiple constraints improved weed optimization (IWO) algorithm is proposed. IWO algorithm separates the multicast request into emergency, common and police requests. We design a multi-objective lion optimization (MOLO) algorithm among fog nodes for resource management, which increases the utilization of resources in fog layer and decrease the response time of multicast session request. MOLO algorithm removes the unnecessary flow table and session table entries in the controller. The DEMR protocol is implemented in Network Simulator (NS3) tool and simulation results are compared with the protocols as EEMSFV, MABC, and CVLMS. From the simulation results we conclude that the DEMR algorithm is better than EEMSFV, MABC and CVLMS in terms of transmission ratio, overhead load, average end to end delay, packet loss ratio and, energy consumption.

Social media monitoring: What can marketers learn from Facebook brand photos?

Users upload >350 million photos per day to Facebook. While considerable research has explored text-based user-generated content on social media, research on photos is still in its early stages. This paper uses a sample of 44,765 Facebook photos from 503 Facebook users in the United States and Germany to determine the degree to which photos play an integral role in people's social media communications. The analysis shows that uploading brand photos (i.e., photos containing a brand name or logo) is related to brand love, brand loyalty, and word-of-mouth (WOM) endorsement of the brand in question. We then code a subsample of these photos for content and train a powerful hybrid machine learning algorithm combining genetic search and artificial neural networks. The resulting algorithm is able to predict users' brand love, brand loyalty, and WOM endorsement from the content of their brand photos posted on Facebook. Finally, we discuss the implications for social media marketing, in particular social media monitoring.

Hybrid Signal Processing and Machine Learning Algorithm for Adaptive Fault Classification of Wind Farm Integrated Transmission Line Protection

The technological advancement in integration of Renewable Green Energy Sources (RGES) like Wind Farm Generators (WFG), and Photovoltaic (PV) system into conventional power system as a future solution to meet the increase in global energy demands in order to reduce the cost of power generation, and improve on the climate change impact. This innovation also introduces challenges in the power system protection by it being compromised due to injected fault current infeeds on existing facilities. These infeed lead to the undesired trip of a healthy section of the line, and protection system failure. This paper presents a soft computational approach to adaptive fault classification model on High Voltage Transmission Line (HVTL) with and without RGES-WFG integration topologies, using extracted one-cycle fault signature of voltage and current signals with wavelet statistical approach in Matlab. The results are unique signatures across all fault types and fault distances with distinct entropy energy values on proposed network architecture. The supervised machine learning algorithm from Bayesian network classified 99.15 % faults correctly with the operation time of 0.01 s to produced best-generalized model with an RMS error value of 0.05 for single line-to-ground (SLG) fault identification and classification. Best suitable for adaptive unit protection scheme integration.

Landslide Susceptibility Assessment by Novel Hybrid Machine Learning Algorithms

Landslides have multidimensional effects on the socioeconomic as well as environmental conditions of the impacted areas. The aim of this study is the spatial prediction of landslide using hybrid machine learning models including bagging (BA), random subspace (RS) and rotation forest (RF) with alternating decision tree (ADTree) as base classifier in the northern part of the Pithoragarh district, Uttarakhand, Himalaya, India. To construct the database, ten conditioning factors and a total of 103 landslide locations with a ratio of 70/30 were used. The significant factors were determined by chi-square attribute evaluation (CSEA) technique. The validity of the hybrid models was assessed by true positive rate (TP Rate), false positive rate (FP Rate), recall (sensitivity), precision, F-measure and area under the receiver operatic characteristic curve (AUC). Results concluded that land cover was the most important factor while curvature had no effect on landslide occurrence in the study area and it was removed from the modelling process. Additionally, results indicated that although all ensemble models enhanced the power prediction of the ADTree classifier (AUCtraining = 0.859; AUCvalidation = 0.813); however, the RS ensemble model (AUCtraining = 0.883; AUCvalidation = 0.842) outperformed and outclassed the RF (AUCtraining = 0.871; AUCvalidation = 0.840), and the BA (AUCtraining = 0.865; AUCvalidation = 0.836) ensemble model. The obtained results would be helpful for recognizing the landslide prone areas in future to better manage and decrease the damage and negative impacts on the environment.

Predicting the daily return direction of the stock market using hybrid machine learning algorithms

Big data analytic techniques associated with machine learning algorithms are playing an increasingly important role in various application fields, including stock market investment. However, few studies have focused on forecasting daily stock market returns, especially when using powerful machine learning techniques, such as deep neural networks (DNNs), to perform the analyses. DNNs employ various deep learning algorithms based on the combination of network structure, activation function, and model parameters, with their performance depending on the format of the data representation. This paper presents a comprehensive big data analytics process to predict the daily return direction of the SPDR S&P 500 ETF (ticker symbol: SPY) based on 60 financial and economic features. DNNs and traditional artificial neural networks (ANNs) are then deployed over the entire preprocessed but untransformed dataset, along with two datasets transformed via principal component analysis (PCA), to predict the daily direction of future stock market index returns. While controlling for overfitting, a pattern for the classification accuracy of the DNNs is detected and demonstrated as the number of the hidden layers increases gradually from 12 to 1000. Moreover, a set of hypothesis testing procedures are implemented on the classification, and the simulation results show that the DNNs using two PCA-represented datasets give significantly higher classification accuracy than those using the entire untransformed dataset, as well as several other hybrid machine learning algorithms. In addition, the trading strategies guided by the DNN classification process based on PCA-represented data perform slightly better than the others tested, including in a comparison against two standard benchmarks.

Image processing based automatic recognition of asphalt pavement patch using a metaheuristic optimized machine learning approach

Patch detection is an important task in pavement condition survey. This study establishes an automatic approach for asphalt pavement patch recognition based on image texture analysis and hybrid machine learning algorithms. Features based on image texture that employs statistical properties of color channels and the gray-scale co-occurrence matrix are used by the Least Squares Support Vector Machine (LSSVM) for discriminating patched areas from non-patch ones. In addition, to optimize the LSSVM training phase, the Differential Flower Pollination (DFP) metaheuristic is used. A data set constructed from a set of 1000 image samples has been utilized to train and verify the proposed integration of image texture analysis techniques, LSSVM, and DFP. Experimental results show that the new model can achieve a good prediction result with Classification Accuracy Rate = 95.30%, Positive Predictive Value = 0.96, and the Negative Predictive Value = 0.95. Additionally, a patch detection program has been developed and compiled in Visual C# .NET to ease the implementation of the hybrid model. Thus, the newly developed method can be a potential tool for traffic management agencies during the phase of pavement condition evaluation.

A hybrid machine learning algorithm for designing quantum experiments

We introduce a hybrid machine learning algorithm for designing quantum optics experiments to produce specific quantum states. Our algorithm successfully found experimental schemes to produce all 5 states we asked it to, including Schrödinger cat states and cubic phase states, all to a fidelity of over 96%. Here, we specifically focus on designing realistic experiments, and hence all of the algorithm’s designs only contain experimental elements that are available with current technology. The core of our algorithm is a genetic algorithm that searches for optimal arrangements of the experimental elements, but to speed up the initial search, we incorporate a neural network that classifies quantum states. The latter is of independent interest, as it quickly learned to accurately classify quantum states given their photon number distributions.

A Comparison of Hybrid Machine Learning Algorithms for the Retrieval of Wheat Biophysical Variables from Sentinel-2

This study focuses on the comparison of hybrid methods of estimation of biophysical variables such as leaf area index (LAI), leaf chlorophyll content (LCC), fraction of absorbed photosynthetically active radiation (FAPAR), fraction of vegetation cover (FVC), and canopy chlorophyll content (CCC) from Sentinel-2 satellite data. Different machine learning algorithms were trained with simulated spectra generated by the physically-based radiative transfer model PROSAIL and subsequently applied to Sentinel-2 reflectance spectra. The algorithms were assessed against a standard operational approach, i.e., the European Space Agency (ESA) Sentinel Application Platform (SNAP) toolbox, based on neural networks. Since kernel-based algorithms have a heavy computational cost when trained with large datasets, an active learning (AL) strategy was explored to try to alleviate this issue. Validation was carried out using ground data from two study sites: one in Shunyi (China) and the other in Maccarese (Italy). In general, the performance of the algorithms was consistent for the two study sites, though a different level of accuracy was found between the two sites, possibly due to slightly different ground sampling protocols and the range and variability of the values of the biophysical variables in the two ground datasets. For LAI estimation, the best ground validation results were obtained for both sites using least squares linear regression (LSLR) and partial least squares regression, with the best performances values of R2 of 0.78, rott mean squared error (RMSE) of 0.68 m2 m−2 and a relative RMSE (RRMSE) of 19.48% obtained in the Maccarese site with LSLR. The best results for LCC were obtained using Random Forest Tree Bagger (RFTB) and Bagging Trees (BagT) with the best performances obtained in Maccarese using RFTB (R2 = 0.26, RMSE = 8.88 μg cm−2, RRMSE = 17.43%). Gaussian Process Regression (GPR) was the best algorithm for all variables only in the cross-validation phase, but not in the ground validation, where it ranked as the best only for FVC in Maccarese (R2 = 0.90, RMSE = 0.08, RRMSE = 9.86%). It was found that the AL strategy was more efficient than the random selection of samples for training the GPR algorithm.

Short-Term Electricity Price Forecasting via Hybrid Backtracking Search Algorithm and ANFIS Approach

In this paper, a hybrid electricity price forecasting method which is composed of two-stage feature selection method and optimized adaptive neuro-fuzzy inference system (ANFIS) technique as a forecasting engine is proposed to accurately forecast electricity price. A multi-objective feature selection approach comprises of multi-objective binary-valued backtracking search algorithm (MOBBSA) as an efficient evolutionary search algorithm and ANFIS method is developed in this paper to extract the most influential subsets of input variables with maximum relevancy and minimum redundancy. Through the combination of backtracking search algorithm (BSA) in learning process of ANFIS approach, a hybrid machine learning algorithm has been developed to forecast the electricity price more accurately. Real-world electricity demand and price dataset from Ontario power market; which is reported as among the most volatile market worldwide, has been used as case study to validate the performance of the proposed approach. From the simulation results, it has been seen that the proposed hybrid forecasting method was effective in accurately forecast the Ontario electricity price. In addition, to prove the superiority of the proposed hybrid forecasting method the simulation results obtained using ANN and ANFIS models optimized by other well-known optimization methods have been compared with that of proposed method.

Mobile Based Healthcare Tool an Integrated Disease Prediction & Recommendation System

Recent advances in mobile technology and machine learning together steer us to create a mobile-based healthcare app for recommending disease. In this study, the authors develop an android-based healthcare app which will detect all kinds of diseases in no time. The authors developed a novel, hybrid machine-learning algorithm in order to provide more accurate results. For the same purpose, the authors have combined two machine-learning algorithms, SVM and GA. The proposed algorithms will enhance the accuracy and at the same time reduce the complexity and count of attributes in the database. Analysis of algorithm is also done using statistical parameters like accuracy, confusion matrix, and roc-curve. The pivotal intent of this research work is to create an android-based healthcare app which will predict disease when provided with certain details. For a disease like cancer, for which a series of tests are required for confirmation, this app will quickly detect cancer and it is helpful to doctors as they can start the right course of treatment right away. Further, this app will also recommend a diet fitting the patient profile.

Forecasting performance comparison of two hybrid machine learning models for cooling load of a large-scale commercial building

The hourly cooling load forecasting of a commercial building is very hard to be guaranteed with high accuracy. Due to the high cost of purchase, installation and maintenance for data acquisition devices in some cases, the forecasting method for univariate time series with nonlinear, random, large fluctuation is investigated in this paper and two hybrid machine learning modelling methods – Chaos–support vector regression (Chaos-SVR) and wavelet decomposition (WD) – SVR are presented. The optimization methods of the lag time and embedding dimension during reconstruction of the phase space are described in detail for Chaos–SVR while the selection processes for the wavelet-based function are also presented for wavelet decomposition –SVR. The prediction accuracy of these two hybrid forecasting methods are compared with backward propagation and SVR according to various evaluation metrics such as the expected error percentage(EEP), mean bias error(MBR), coefficient of variation of root mean square deviation(CV-RMSE), mean absolute percentage error (MAPE) and R2 determination coefficient. The forecasting results show that the Chaos–SVR modelling method outperforms the WD-SVR while EEP of Chaos-SVR is 7.4% better than WD-SVR– because of the chaotic characteristics of the cooling load for commercial buildings, and both of the hybrid algorithms outperform the single prediction algorithms, BP and SVR whose EEPs are at least 60% better than that of the other single ones. Although hybrid machine learning algorithms involve more complex modelling procedures than single prediction algorithms, their prediction times are still less than the single ones. The hybrid forecasting methods proposed in this work could also be used in other forecasting fields.

Climate Change and Power Security: Power Load Prediction for Rural Electrical Microgrids Using Long Short Term Memory and Artificial Neural Networks

Many of rural Alaskan communities operate their own, stand-alone electrical microgrids as there is no state-wide power distribution network. Although the fossil fuel-based power generators are the primary energy source in these isolated communities, an increasing number of microgrids have started to diversify their power supply by including renewable wind, hydro and photovoltaic energy sources. The integration and optimization of the multiple energy sources requires a design of a new power management system that can anticipate how much electricity will be drawn by the community and how much electricity will be generated from the renewable sources in order to control the fossil fuel-based generators to meet the community power demand. To address this problem, we designed a hybrid machine learning algorithm to predict community power draw as one module for the next generation microgrid power management system. The algorithm divides the task of a power load prediction into two sub-models: the first temporal model predicts the future weather conditions and the second model is trained to associate the predicted weather conditions with the community power demand. The results illustrate (1) the feasibility of building a machine learning algorithm that uses relatively small data for model training and validation, (2) the ability to predict the near-future community power load for the microgrids operating in the environments with highly dynamic weather conditions and (3) how the integration of multiple low quality future weather conditions produced high accuracy power load prediction.

Using human brain activity to guide machine learning

Machine learning is a field of computer science that builds algorithms that learn. In many cases, machine learning algorithms are used to recreate a human ability like adding a caption to a photo, driving a car, or playing a game. While the human brain has long served as a source of inspiration for machine learning, little effort has been made to directly use data collected from working brains as a guide for machine learning algorithms. Here we demonstrate a new paradigm of “neurally-weighted” machine learning, which takes fMRI measurements of human brain activity from subjects viewing images, and infuses these data into the training process of an object recognition learning algorithm to make it more consistent with the human brain. After training, these neurally-weighted classifiers are able to classify images without requiring any additional neural data. We show that our neural-weighting approach can lead to large performance gains when used with traditional machine vision features, as well as to significant improvements with already high-performing convolutional neural network features. The effectiveness of this approach points to a path forward for a new class of hybrid machine learning algorithms which take both inspiration and direct constraints from neuronal data.

Hybrid machine learning algorithms for fault detection in android smartphones

AbstractThe newest trends with smart devices and the concept of Internet of Things (IoT) is that of big data and analytics. More recently, the concept of mobile cloud computing has emerged with advances in terms of hardware and communication speeds. However, there are serious implications on the normal lifetime of the battery‐powered smart devices being used for these different applications. Most sensors deployed for different smart IoT applications are already available in smart phones and used by the majority of individuals on a daily basis. With the recent explosions of the Samsung Galaxy Note 7 smartphones, safety of individuals has climbed the priority list in the IoT field. Along this line, it is important to address fault detection and prediction of smartphones, which are being used in fields such as mobile health. In this paper, a cloud‐based open‐source framework for capturing and processing real‐time streaming information (eg, screen brightness, CPU usage, battery level, voltage, device temperature, and Wi‐Fi signal strength) about different android‐based smartphones has been set up. A NoSQL Cassandra database is used and a Spark distributed computing Scala‐based framework accesses the data for further processing. Classification machine learning algorithms (Naïve‐Bayes (NB), decision tree (DT), and random forest) are used to obtain trained models for predicting symptoms of faulty behaviors in smartphones. The performance of a hybrid 2‐staged machine learning mechanism is proposed whereby cascaded classification algorithms are used. Results show that when using DT algorithm (Level 1) and NB algorithm (Level 2), a slight increase in percentage accuracy is observed. This demonstrates that there is the possibility for further improving the NB model for classification and fault detection when using a combined training model with DT algorithm (Level 1) and NB algorithm (Level 2). The results for the cascaded model show that the percentage accuracy of most hybrid models in this case are slightly inferior to the standalone models themselves. The hybrid models using the DT algorithm at both levels and DT algorithm followed by RF algorithm do not improve the percentage accuracy. However, this demonstrates that the robustness of the classification algorithm can be maintained with this type of hybrid classification algorithm and leaves room for further research of possible techniques and ways to improve the accuracy.

Short term solar insolation prediction: P-ELM approach

The accurate forecasting of solar irradiance with hybrid machine learning algorithm is presented in this paper. A novel Persistence-Extreme Learning Machine (P-ELM) algorithm is used for training of the system. The Clearness Index (CI) value of the 22 districts of Andhra Pradesh (India) is calculated and out of which four areas are identified with highest CI values. The Global Horizontal Irradiance (GHI) is forecasted for the selected areas with different weather conditions such as winter, summer and rainfall seasons using a P-ELM algorithm. The input parameters are Temperature, Diffuse horizontal irradiance, pressure and past GHI and GHI for the next instant as the output is considered. The real time data is obtained for every one hour interval for a period of one month. The performance of the P-ELM algorithm is evaluated in terms of Mean Absolute Error and Root Mean Square Error. From the obtained results, it is observed that P-ELM algorithm offers better performance over the fundamental P-ELMs. The P-ELM algorithm gives good forecasting accuracy with minimum simulation time. The simulation of P-ELM algorithm is carried out using MATLAB 2013a environment. The P-ELM algorithm is very much beneficial for accurate and reliable real time solar forecasting.

A hybrid machine learning approach for additive manufacturing design feature recommendation

PurposeThis paper aims to present a hybrid machine learning algorithm for additive manufacturing (AM) design feature recommendation during the conceptual design phase.Design/methodology/approachIn the proposed hybrid machine learning algorithm, hierarchical clustering is performed on coded AM design features and target components, resulting in a dendrogram. Existing industrial application examples are used to train a supervised classifier that determines the final sub-cluster within the dendrogram containing the recommended AM design features.FindingsThrough a case study of designing additive manufactured R/C car components, the proposed hybrid machine learning method was proven useful in providing feasible conceptual design solutions for inexperienced designers by recommending appropriate AM design features.Originality/valueThe proposed method helps inexperienced designers who are newly exposed to AM capabilities explore and utilize AM design knowledge computationally.

Data mining and machine learning for identifying sweet spots in shale reservoirs

Due to its complex structure, production form a shale-gas formation requires more drillings than those for the traditional reservoirs. Modeling of such reservoirs and making predictions for their production also require highly extensive datasets. Both are very costly. In-situ measurements, such as well-logging, are one of most indispensable tools for providing considerable amount of information and data for such unconventional reservoirs. Production from shale reservoirs involves the so-called fracking, i.e. injection of water and chemicals into the formation in order to open up flow paths for the hydrocarbons. The measurements and any other types of data are utilized for making critical decisions regarding development of a potential shale reservoir, as it requires hundreds of millions of dollar initial investment. The questions that must be addressed include, does the region under study can be used economically for producing hydrocarbons? If the response to the first question is affirmative, then, where are the best places to carry out hydro-fracking? Through the answers to such questions one identifies the sweet spots of shale reservoirs, which are the regions that contain high total organic carbon (TOC) and brittle rocks that can be fractured. In this paper, two methods from data mining and machine learning techniques are used to aid identifying such regions. The first method is based on a stepwise algorithm that determines the best combination of the variables (well-log data) to predict the target parameters. However, in order to incorporate more training, and efficiently use the available datasets, a hybrid machine-learning algorithm is also presented that models more accurately the complex spatial correlations between the input and target parameters. Then, statistical comparisons between the estimated variables and the available data are made, which indicate very good agreement between the two. The proposed model can be used effectively to estimate the probability of targeting the sweet spots. In the light of an automatic input and parameter selection, the algorithm does not require any further adjustment and can continuously evaluate the target parameters, as more data become available. Furthermore, the method is able to optimally identify the necessary logs that must be run, which significantly reduces data acquisition operations.

Validation of natural language processing (NLP) for automated ascertainment of EGFR and ALK tests in SEER cases of non-small cell lung cancer (NSCLC).

6528 Background: The Surveillance, Epidemiology, and End Results (SEER) registries lack information on the Epidermal Growth Factor Receptor (EGFR) mutation and Anaplastic Lymphoma Kinase (ALK) gene rearrangement test results. With the goal of enabling population-based outcomes research in molecularly selected NSCLC subgroups, we conducted a validation study of NLP for ascertainment of EGFR and ALK testing from electronic pathology reports (e-paths) of patients included in the Seattle-Puget Sound (SPS) and Kentucky Cancer (KCR) SEER registries. Methods: We obtained 4,278 and 1,041 e-paths pertaining to 1,634 and 565 patients with stage IV non-squamous NSCLC diagnosed from 1/1/2011 to 12/31/2013 and included in the SPS and KCR registries, respectively. Two oncologists independently reviewed all reports to generate a gold-standard dataset. We used 855 of the SPS reports to train hybrid rule-based and machine learning algorithms for detection of test status (reported vs. not reported), and test result if reported (positive vs negative) for EGFR mutational analysis and ALK testing by FISH, IHC, or gene sequencing. In the remaining 3,423 SPS reports, we conducted a 5-fold cross-validation analysis to estimate the internal NLP sensitivity, specificity, positive predictive value, and negative predictive value for test status and results, respectively. We used a hierarchical rules system to assess the NLP accuracy at the patient level. For external validation, we repeated all analyses in the KCR dataset. Results: In the SPS internal validation report sample, the validity metrics ranged from 97% to 99% for EGFR and ALK test status, and from 95% to 100% for EGFR and ALK test results, respectively. In the KCR external validation report sample, the metrics ranged from 74% to 96% for EGFR and ALK test status, and 2% to 100% for test results, respectively. At the patient level, the NLP accuracy for EGFR and ALK was 95% and 96% (SPS cohort), and 70% and 72% (KCR cohort) respectively. Conclusions: NLP is a valid method for determining EGFR and ALK test status and results for patients included in SEER registries with access to e-path, but the algorithms likely need to be registry-specific.

Improve the Performance of a Complex FMS with a Hybrid Machine Learning Algorithm

Modern manufacturing systems are expected to undertake multiple tasks, flexible for extensive customization, and that trends make production systems become more and more complicated. The advantage of a complex production system is a capability to fulfill more intensive goods production and to adapt to various parameters in different conditions. The disadvantage of a complex system, on the other hand, with the pace of the increase of complexity, lies in the control difficulties rising dramatically. Moreover, classical methods are reluctant to control a complex system, and searching for the appropriate control policy tends to become more complicated. Thanks to the development of machine learning technology, this problem is provided with more possibilities for the solutions. In this paper, a hybrid machine learning algorithm, integrating genetic algorithm and reinforcement learning algorithm, is proposed to cope with the accuracy of a control policy and system optimization issue in the simulation of a complex manufacturing system. The objective of this paper is to cut down the makespan and the due date in the manufacturing system. Three use cases, based on the different recipe of the product, are employed to validate the algorithm, and the results prove the applicability of the hybrid algorithm. Besides that, some additionally obtained results are beneficial to find out a solution for the complex system optimization and manufacturing system structure transformation.