Asia-Pacific Forum on Science Learning and Teaching, Volume 19, Issue 1, Article 8 (Jun., 2018)
Ananta Kumar JENA
Predicting learning outputs and retention through neural network artificial intelligence in photosynthesis, transpiration and translocation

Previous Contents Next


Methodology

Pretest-posttest single group experimental design used to investigate the existing misconception status of the participants. However, the improvement of learning performance and retention was assessed after exposed to neural network artificial intelligence for its broader generalization. Out of forty students of a 8th standard, five students’ age ranged 13.5-14.6 years and the mean age was 14.5 with S.D= 9.8 found 60% to 80% of misconception in photosynthesis, transpiration and translocation test. Hierarchical multiple regression model assessed the relationship between more than two independent variables and a dependent variable in the present study. The output contributes important information. a)R is the correlation between the predicted values and the observed values of Y, and b) R-square indicates how well a regression model predicts responses for new observations (R2 represents the proportion of variance in the dependent variable explain by the independent variable, in the regression model, which is beyond, and below the mean mode). c) An adjusted R2 is the value corrects positive biased of the population, d) F value, and the degree of freedom shows the statistical significance of the regression model;  and e) the co- efficient (β) is a constant, and independent variable predicts the dependent variable. The detail of the design of the study is given in table 1.

Table 1-Design of the Study*

Group

Pre-test

Treatment

Post-test

Delayed-test

Experimental

x
NNAI
x
x

Instrumentation

Misconception test cum Achievement test

Jena (2015) developed a misconception test cum achievement test in photosynthesis, transpiration and translocation based on the syllabus for standard 8th students affiliated to NCERT, New Delhi. It has thirty multiple-choice items having four point options where the normative group for the Misconception test cum Achievement test randomized among the cross-cultural group of Indian who accurately reflected the diversity of that group of respondents of the test. In psychology, the normative group for the misconception test used among the students of 13.5-14.6 year from various demographic groups in India. The Content Validity Ratio (CVR=8.0) and Cronbach alpha was .86 (see Appendix I).

Neural Network Blank sheet

Through the misconception test it was identified that five students who had 60%-80% of misconceptions in the photosynthesis, transpiration, and translocation concepts.  Before experiment, the researcher provided two tutorial classes to acquaint with the neural network teaching. Neural network is an online software developed by University of Missouri, Columbia is available in http://www.semanticresearch.com. By the help of this software, students could construct their mental and cognitive information on the concept of photosynthesis, transpiration, and translocation. Students can be directly followed the steps like: selection of neurons(inputs), link all neurons with the hidden neurons, link neurons with hidden neurons, making links with hidden neuron and outputs, and sharing of individual neural network with peers. In this study, neural networks used when the exact nature of the relationship between inputs and outputs were unknown. A key feature of neural networks training was to train the students on how to link inputs and outputs. There were three types of training in neural networks was used as networks supervised, unsupervised training and reinforcement learning with supervised. Some neural network training techniques were back propagation, quick propagation, conjugate gradient descent, projection operation and some unsupervised neural networks are multi layer. Artificial Neural Networks (ANN) has input, scaled input, hidden neurons, scaled output, and output (https:// www. xenonstack.com/blog/overview-of-artificial-neural-networks-and-its-applications). Input and scaled inputs belong to input layer and hidden neurons are hidden layers. The schematic representation of an artificial neural network is given in Fig 1. Input layer contains the units of artificial neurons which receive input from the outside world on which network will learn, recognize about photosynthesis, transpiration and translocation or other process. It means all the concepts, sub concepts, micro concepts, and examples students will learn from the environment. Output layer contains units that respond to the information about how it helps to learn any task. Hidden layer are the units are in between input and output layers. The job of hidden layer is to transform the input into something that output unit can use in some way.

Figure 1. Artificial Neural Network Blank Sheet

Input layer contains the units of artificial neurons which receive input from the outside world on which network will learn, recognize about photosynthesis, transpiration and translocation or other process. It means all the concepts, sub concepts, micro concepts, and examples students will learn from the environment. Output layer contains units that respond to the information about how it helps to learn any task. Hidden layer are the units are in between input and output layers. The job of hidden layer is to transform the input into something that output unit can use in some way.

Learning Techniques in Neural Networks

Before instruction, a misconception test on photosynthesis, transpiration, and translocation administered to the participants, and after scoring the answer sheets, it showed that 60 -80% of concepts were misconceived. The participants were advised to use online and offline neural network blank sheets on photosynthesis, transpiration and translocation concepts (Appendix-I). Neural network template used to frame the concepts of photosynthesis, transpiration, and translocation in input & hidden positions.  In fact, a twenty-contact hour of intervention was provided to learn and practice the propositional neural network on photosynthesis, transpiration, and translocation concepts. Every day, thirty minutes, students used internet to develop neural network and followed by that thirty minutes they practiced through offline neural network blank sheet. However, the learners felt comfortable with the offline mode neural network blank sheet to conceptualize photosynthesis, transpiration, translocation rather than online. This instruction cum self-learning practice continued up to twenty contact hours. After instruction, the achievement test (the earlier misconception test) was administered to the students as the posttest and the same test was administered after one month was the delayed test used to assess the retention. To predict the actual output, students were tested through Neural Network Blank sheet. Both pretest and posttest score analyzed to predict the students’ output or performance, through input layer and hidden layers. The Neural Network Blank sheets were scored in such a manner that for each correct output 10 points, and for each partial output 5 points and these points or marks were provided to the students according to their number input layer, in different hidden layers and output layer. Finally, linear regression analysis used to predict the output performance from the input and hidden layer entries. SPSS version 21 used to identify the predictors of Photosynthesis, transpiration, and translocation. The details of the activity of NNAI on Photosynthesis, Transpiration, and Translocation are showing in Figures 2, 3, & 4.

Figure 2. Neural network artificial intelligence on Photosynthesis

Figure 3. Neural network artificial intelligence on Transpiration

Figure 4. Neural network artificial intelligence on Translocation

 


Copyright (C) 2018 EdUHK APFSLT. Volume 19, Issue 1, Article 8 (Jun., 2018). All Rights Reserved.