Approach of hybrid soft computing for agricultural data classification

Soft computing is an important computational paradigm,and it provides the capability of flexible information processing to solve real world problems.Agricultural data classification is one of the important applications of computing technologies in agriculture,and it has become a hot topic because of the enormous growth of agricultural data available.Support vector machine is a powerful soft computing technique and it realizes the idea of structural risk minimization principle to find a partition hyperplane that can satisfy the class requirement.Rough set theory is another famous soft computing technique to deal with vague and uncertain data.Ensemble learning is an effective method to learn multiple learners and combine their decisions for achieving much higher prediction accuracy.In this study,the support vector machine,rough set and ensemble learning were incorporated to construct a hybrid soft computing approach to classify the agricultural data.An experimental evaluation of different methods was conducted on public agricultural datasets.The experimental results indicated that the proposed algorithm improves the performance of classification effectively.

Research on the Relationship Between Garlic and Young GarlicShoot Based on Big Data

In view of the problems such as frequent fluctuation of garlic price, lack ofefficient forecasting means and difficulty in realizing the steady development of garlicindustry, combined with the current situation of garlic industry and the collected datainformation. Taking Big Data platform of garlic industry chain as the core, using themethods of correlation analysis, smoothness test, co-integration test, and Grangercausality test, this paper analyzes the correlation, dynamic, and causality between garlicprice and young garlic shoot price. According to the current situation of garlic industry,the garlic industry service based on Big Data is put forward. It is concluded that there is apositive correlation between garlic price and young garlic shoot price, and there is a longtermstable dynamic equilibrium relationship between young garlic shoot price and garlicprice fluctuation, and young garlic shoot price can affect garlic price. Finally, it isproposed to strengthen the infrastructure construction of garlic Big Data, increase thetechnological innovation and application of garlic Big Data technology, and promote thesafety and security ability of the whole industry to promote the development of garlicindustry.

Thematic harvesting of agricultural resources from generic repositories

Metadata aggregators and service providers harvest entire collections or they restrict harvesting by date or sets.However most often user approach to collections is not by dates or set names but by domain based keywords.Harvesting by domains is an issue when service providers attempt to collect data from multiple sources.The main problem is that harvesters,at present,do not have the facility to distinguish themes such as domains.In the present work,an attempt has been through Tharvest,a thematic harvester model using the proposed methodology harvesting agricultural resources from generic repositories.Tharvest encompasses a process where technical terms of the domain of agriculture are taken from AGROVOC,a multilingual,structured controlled vocabulary designed to cover concepts and terminologies in the agriculture domain.AGROVOC is deployed to provide the basis for selective harvesting.The system components and workflows are presented and described.Metadata aggregators provide end-users a single platform discovery facility to resources collected from various data providers.It is observed that aggregators such as INDUS[www.drtc.isibang/ac.in/indus]dealing with agriculture and related domains facilitate aggregating metadata from not only repositories but also other sources such as journals and enable a centralized access to full text and objects.While harvesting can be fairly simple and straight forward,it is not without its challenges.This paper intends to highlight some of the issues in harvesting metadata in agricultural domain.The particular focus is to identify agriculture related metadata from generic sets.

Yield performance estimation of corn hybrids using machine learning algorithms

Estimation of yield performance for crop products is a topic of interest in agriculture.In breeding programs,we cannot test all possible hybrids created by crossing two parents(inbred and tester)since it would be too time consuming and costly.In this paper,we exploit different machine learning algorithms including decision tree,gradient boosting machine,random forest,adaptive boosting,XGBoost and neural network to predict the yield of corn hybrids using data provided in the 2020 Syngenta Crop Challenge.The participants were asked to predict the yield of missing hybrids which were not tested before.Our results show that the prediction obtained by XGBoost is more accurate than other models with a root mean square error equal to 0.0524.Therefore,we use XGBoost model to estimate the yield performance for untested combinations of inbreds and testers.Using this approach,we identify hybrids with high predicted yield that can be bred to increase corn production.

Scientific development of smart farming technologies and their application in Brazil

Smart farming(SF)involves the incorporation of information and communication technologies into machinery,equipment,and sensors for use in agricultural production systems.New technologies such as the internet of things and cloud computing are expected to advance this development,introducing more robots and artificial intelligence into farming.Therefore,the aims of this paper are twofold:(i)to characterize the scientific knowledge about SF that is available in the worldwide scientific literature based on the main factors of development by country and over time and(ii)to describe current SF prospects in Brazil from the perspective of experts in this field.The research involved conducting semi-structured interviews with market and researcher experts in Brazil and using a bibliometric survey by means of data mining software.Integration between the different available systems on the market was identified as one of the main limiting factors to SF evolution.Another limiting factor is the education,ability,and skills of farmers to understand and handle SF tools.These limitations revealed a market opportunity for enterprises to explore and help solve these problems,and science can contribute to this process.China,the United States,South Korea,Germany,and Japan contribute the largest number of scientific studies to the field.Countries that invest more in R&D generate the most publications;this could indicate which countries will be leaders in smart farming.The use of both research methods in a complementary manner allowed to understand how science frame the SF and the mains barriers to adopt it in Brazil.