In vitro micropropagation of Chlorophytum borivilianum: A Predictive

Model Employing Artificial Neural Networks trained with a range of

Algorithms

PREETI KAUSHIK

Department of Biotechnology, University Institute of Engineering and Technology, Maharshi

Dayanand University, Rohtak, Haryana, INDIA

NEHA KHURANA

Department of Electrical Engineering, University Institute of Engineering and Technology, Maharshi

Dayanand University, Rohtak, Haryana, INDIA

MADHU RANI

Department of Biotechnology, University Institute of Engineering and Technology, Maharshi

Dayanand University, Rohtak, Haryana, INDIA

GOPAL KRISHAN

IIMT, Greater Noida, UP

SONIA KAPOOR*

Department of Biotechnology, University Institute of Engineering and Technology, Maharshi

Dayanand University, Rohtak, Haryana, INDIA

*Corresponding Author

Abstract: The formulation of plant tissue culture media continues to be a complex undertaking, primarily due to

the intricate interplay of multiple components. Numerous factors (such as genotype, disinfectants, media pH,

temperature, light, and immersion time) interact to affect the process of plant tissue culture. The artificial neural

network is considered one of the most potent computational techniques that has emerged as a highly potent and

valuable methodology for effectively representing intricate non-linear systems. This research paper focuses on

the development of a predictive model for determining the number of shoots in response to different

macronutrient compositions in the culture medium used for in-vitro micropropagation of Chlorophytum

borivilianum. The study employs artificial neural networks (ANNs) trained with different algorithms to

accurately predict the number of shoots and shoot length of the plant species. These algorithms include the

Levenberg-Marquardt (LM), Scaled Conjugate Gradient (SCG), and Bayesian Regularisation (BR)

backpropagation algorithms. A feed-forward backpropagation network was constructed with a single hidden layer

consisting of ten nodes and two output units in the output layer. The input vector contained five elements. The

transfer functions 'tansig' and 'purelin' were utilized for the hidden and output layers, respectively. In this study,

the effectiveness of neural networks was tested by contrasting the outcomes with real-life data gathered from in-

depth tissue culture experiments, which was named the target set. The comparative analysis of "Mean Square

Error" and Pearson's correlation coefficient (R) were used to evaluate the effectiveness of networks for improved

training initialization. The prediction ability of Levenberg-Marquardt was found superior to other training

algorithms with an R-value of 9.92 also the output range of network ‘trainlm’ was closest to the empirical target

range during the comparison of experimental target data ranges from wet lab practice.

Keywords: Artificial neural network, Bayesian Regularisation, Chlorophytum borivilianum, Levenberg-

Marquardt, Scaled Conjugate Gradient (SCG)

Received: April 23, 2022. Revised: February 9, 2023. Accepted: March 11, 2023. Published: April 10, 2023.

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

12

Volume 2, 2023

1. Introduction

In vitro micropropagation, a widely employed

technique for plant propagation involves cultivating

plant tissues in a controlled environment. The

success of this technique largely depends on the

composition of the culture medium, particularly the

macronutrient content. Accurate prediction of the

macronutrient composition is crucial for achieving

optimal growth and development of the target plant

species [1,22]. The majority of medicinal plants are

going to be extinct if no steps are taken to preserve

them. Chlorophytum borivilianum is a plant that is

the epicentre of various medicinal properties and

needs to be preserved with the boon of invitro

micropropagation. A multilayer perceptron model

with a feed-forward architecture has been applied to

predict the shoot biomass of the mentioned plant

[1,2].

Artificial neural networks (ANNs) have emerged as

a prominent machine learning (ML) technique for

the purpose of modeling and predicting intricate

processes. ANNs are widely regarded as one of the

most potent tools in the field, owing to their ability

to effectively capture and represent complex

patterns and relationships within data. They are

computer simulations with biological inspiration

that are used to carry out particular tasks. Neural

networks are commonly shown as interconnected

systems of "neurons" capable of performing

computations by transmitting information through

them. Artificial Neural Networks (ANNs) can be

effectively utilized in machining processes to predict

response parameters based on process parameters,

provided that they have been appropriately trained.

It is imperative to exercise caution and adhere to

proper protocols while implementing Artificial

Neural Networks (ANN) in these processes [3]. The

design and implementation of an ANN must ensure

that the set of input data leads to the intended output

(either directly or through the use of a relaxation

process). Various techniques can be employed to

measure the magnitude of the connections. In other

words, the weights can be predetermined based on

prior information, or the neural network can be

trained by inputting learning patterns and allowing

the network to modify the weights based on a

specified learning rule. An artificial neural network

can be trained using supervised learning as well as

unsupervised learning methods. Though there are

other learning methods present, such as

reinforcement learning methods, we will focus on

supervised learning methods in this paper. The

primary objective of a supervised learning algorithm

is to ascertain a mapping function that effectively

relates the input variable (x) to the output variable

(y). There exist numerous types of artificial neural

network (ANN) methods, which encompass

perceptron, backpropagation, Learning vector

quantization (LVQ), probabilistic neural network,

Hopfield, and radial base network [4, 23]. The Back

Propagation (BP) algorithm, however, is the most

well-known and often employed learning technique

for estimating the values of the weights. The

Backpropagation Neural Network (BPNN) is a

feedforward neural network utilizing chain rules.

This approach utilizes a predetermined set of input

and output values to determine the optimal weight

and bias parameters for the neural network [5, 6].

The learning process consists of two distinct stages:

the forward transmission of signals and the

backward propagation of errors. The method

terminates when the error function's value becomes

negligibly small. The process chronology of the

Backpropagation algorithm is shown in Figure 1.

Traditional BP networks, however, have certain

drawbacks, including slow convergence and a

simple fall to the local minimum [7]. To mitigate the

error associated with the backpropagation algorithm,

various generalization methods have been utilized.

These methods include Bayesian regularisation

(BR) [8], Levenberg–Marquardt (LM) [9], and

Scaled conjugate gradient [10]. These methods have

been chosen due to their ability to achieve a lower

mean squared error.

The Levenberg–Marquardt algorithm, also known as

the LM algorithm, is a numerical optimization

method that was independently developed by

Kenneth Levenberg and Donald Marquardt. This

algorithm is specifically designed to solve the

problem of minimizing a nonlinear function. The

observed phenomenon exhibits rapidity in its

execution and demonstrates a consistent pattern of

convergence [24]. This approach can be used to train

small- and medium-sized issues in the field of

artificial neural networks. In convergent situations,

the Levenberg-Marquardt algorithm converges far

more quickly than the steepest descent method,

although it tends to be slightly slower than the

Gauss-Newton algorithm. The fundamental concept

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

13

Volume 2, 2023

underlying the Levenberg-Marquardt algorithm is its

utilization of a combined training process [11].

Bayesian regularisation is a novel approach that

combines the principles of Bayesian methods and

Artificial Neural Networks (ANN) to automatically

determine the optimal regularisation parameters.

This technique aims to enhance the performance and

generalization capabilities of ANN models by

incorporating Bayesian principles into the

regularisation process[26]. By leveraging Bayesian

regularisation, researchers and practitioners can

effectively address the challenge of selecting

appropriate regularisation parameters, which is a

critical task in training ANN models. In contrast to

the conventional methodology employed in network

training, which entails selecting the optimal weight

set by minimizing the error function, the Bayesian

approach adopts a different perspective by

considering the probability distribution of network

weights. In light of the aforementioned observations,

it can be deduced that the predictions generated by

the network exhibit characteristics analogous to a

probability distribution [12]. Another approach

known as the scaled conjugate gradient (SCG),

which was introduced by Moller in 1993, utilizes

conjugate directions. However, unlike previous

conjugate gradient algorithms that necessitate a line

search at each iteration, the SCG algorithm does not.

The SCG algorithm was developed to circumvent

the laborious process of line search. It utilizes a step

size that is determined by a quadratic approximation

of the error function. This characteristic enhances

the algorithm's robustness and reduces its

dependence on user-defined parameters [13, 25].

Fig 1: Several distinct stages comprise the

procedure for supervised learning

The majority of medicinal plants are going to be

extinct if no steps are taken to preserve them.

Chlorophytum borivilianum is a plant that is the

epicenter of various medicinal properties and needs

to be preserved with the boon of invitro

micropropagation [14]. A multilayer perceptron

model with a feed-forward architecture has been

applied to predict the shoot biomass of the

mentioned plant. The primary aim of this research

paper is to provide a thorough comparative analysis

of the utilization of several artificial neural network

training techniques in the prediction of no. of shoots

and shoot length. This study investigates three

training strategies for a multilayer perceptron (MLP)

feedforward neural network: Levenberg-Marquardt

(LM), Scaled Conjugate Gradient (SCG), and

Bayesian Regularisation (BR) backpropagation

procedures. The evaluation of the performance is

conducted using various statistical metrics,

including the mean square error (MSE), and

Pearson's correlation coefficient (r). MATLAB is

selected as the testing software for this purpose to

carry out the necessary computations and

visualizations.

2. Material and Methods

2.1 Compilation of Input data

The data used in this study was compiled after

invitro propagation of chlorophytum borivilianum

nodal explants grown in a Murashige and Skoog

(MS) medium with different combinations of

macronutrient concentrations ranging from 0.5 mg/l

to 2.25 mg/l supplemented with 3% sucrose and

0.8% agar. All the explants were grown in a

controlled environment of a 16h light/8h dark

photoperiod at 25±2°C. The shoot length and

number of shoots were noted after 20 days of culture

establishment. The data retrieved from the

experiments was used to train the Multilayer

perceptron models.

2.2 Training the ANN models

Neural networks can be categorized based on their

intended purpose, as well as their fundamental

topology and the training method employed. The

automation of commercial mass propagation of

plants relies heavily on decision-making networks,

which fall into the classification and clustering

Determine

training

dataset

type

Get labelled

data for

training

Divide the

training

dataset into

test,

validation,

and

training.

Choose an

algorithm

for the

model

Use the

training

dataset to

run the

algorithm

Provide the

test set to

assess

model

accuracy.

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

14

Volume 2, 2023

models of neural networks [15]. The primary

objective of this study is to evaluate the

effectiveness of different training algorithms for

neural networks (NN) in the field of plant tissue

culture research. Among the different techniques

available for neural networks, the specific approach

employed in this study was the 'Backpropagation'

(BPN) technique trained with Levenberg-Marquardt

(LM), Scaled Conjugate Gradient (SCG), and

Bayesian Regularisation (BR) algorithms.

Fig 2: Artificial neural network model architecture

Feed-forward backpropagation-type network

architectures were developed. Each network

consisted of an input layer with five input nodes, a

single hidden layer with ten nodes, and one output

layer with two nodes trained with Levenberg-

Marquardt (trainlm), Scaled Conjugate Gradient

(trainscg), and Bayesian Regularisation (trainbr)

functions, as shown in figure 2. The inputs to the

ANN model were the varying concentrations of five

media components, i.e., Ammonium nitrate,

Potassium nitrate, Calcium chloride anhydrous,

Magnesium sulfate, and Potassium phosphate

monobasic. 42 combinations of the macronutrients

were used to train the data. The datasets are

categorized into three distinct subsets, namely the

training set, validation set, and testing set. An altered

transfer function between the hidden layer ('tansig')

and the output layer ('purelin') has been

implemented. Both the weight and the bias were

initially set at random and were changed as the

model was trained. It has been trained using a total

of 250 epochs. The sigmoid function is used by the

neuron set to calculate the weighted sum of its

inputs. In the context of network analysis, the

computed result is compared with the expected

result to determine the accuracy of the network's

performance. This comparison allows for the

quantification of the discrepancy between the

computed and expected results, which is commonly

referred to as the error on the network. The error

value is subsequently employed during the

backward propagation step to modify the weights of

neurons [16]. The output neuron generates the net

output from the input neurons. Results from this

study were compared with empirical data from

thorough tissue culture experiments conducted to

optimize different growth factors, allowing for a

definitive conclusion to be drawn about the efficacy

of neural networks.

3. Results and Discussion

The primary goal of this research was to evaluate the

accuracy with which various backpropagation

training algorithms predicted both the number of

shoots and the length of shoots produced by in vitro

propagation of Chlorophytum borivilianum. The

evaluation of network efficiency for improved

training initialization can be assessed through a

comparative analysis of Mean Squared Error (MSE)

and Pearson's correlation coefficient (r) for each

trained network tabulated in Table 1.

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

15

Volume 2, 2023

R(Pearson’s correlation

coefficient )

9.92

9.189

9.81

Table1: Compilation of Mean Square Error and Pearson's correlation coefficient predicted by different

algorithms

The Pearson correlation coefficient (r) is widely

utilized as a primary method for quantifying the

strength and direction of a linear relationship. The

correlation coefficient is a statistical metric that

quantifies the magnitude and direction of the

association between two variables, with values

ranging from -1 to 1[17]. The Mean Squared Error

(MSE) quantifies the discrepancy between the

estimated or predictive model and the observed

values within a particular sample. The metric

calculates the mean squared deviation between the

anticipated values and the actual values, quantifying

the disparity between the model's predictions and the

observed data. The Mean Squared Error (MSE) is

commonly employed as a metric for evaluating the

performance of a model by comparing its predictions

on the complete training dataset to the actual label or

output value [18]. Accordingly, the efﬁciency of the

network for training initialization including the

training function was trainlm> trainbr >trainscg. The

efficiency of trained networks for the least deviation

from the target range was assessed by comparative

observation between the experimental target data

range and all trained network output ranges. Though

the predictions made by all the MLPs trained with

different algorithms were similar, the result

predicted by the Levenberg-Marquardt algorithm

was closer to the actual experimental results.

Fig 3: Regression plots as an output of Levenberg-Marquardt, Scaled Conjugate Gradient, and Bayesian

Regularisation training algorithms

Hence, in the present investigation conducted on

trained and tested datasets, it was observed that the

trained network utilizing the “trainlm” transfer

function, along with a three-layer feed-forward back

propagation methodology, demonstrated proficiency

in predicting optimal values for significant physical

Results

Levenberg-

Marquardt

Scaled Conjugate

Gradient

Bayesian

Regularisation

MSE

1.82

1.91

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

16

Volume 2, 2023

parameters in the context of upscale culture. These

parameters have a direct impact on biomass growth

and, consequently, enhance overall productivity. The

predicted values obtained in this study exhibit

comparability to the values derived from

experimental investigations as shown in Table 2. In

conclusion, this approach offers a straightforward

and unbiased means of making predictions in this

domain and presents a neural network-based

methodology for predicting the growth and

productivity of in vitro cultured plants which can be

used for successfully scaling up the cultivation

process in larger bioreactors. Furthermore, neural

networks have the potential to be utilized in the field

of plant tissue culture in conjunction with other

approaches to enhance the accuracy of results. These

techniques may include image analysis, multiple

regression modeling, and computer programming

[19].

S.n

o.

Inputs

Outputs

Ammoni

um

nitrate

Potassi

um

nitrate

Calciu

m

chlorid

e

anhydr

ous

Magnesi

um

sulphate

Potassi

um

phosph

ate

monob

asic

Experime

ntal

Output

Bayesian

Regularisa

tion

Levenb

erg

Marqua

rdt

algorith

m

Scaled

conjug

ate

1.

1.42

11.86

11.897

11.859

11.197

2.

1.42

0.58

1.42

13.43

12.730

13.430

12.880

3.

0.58

1.42

0.58

12.47

11.168

12.469

12.448

4.

1

2

1

13.34

13.186

13.339

13.680

5.

1.42

0.58

1.42

10.24

10.060

10.240

9.636

6.

1.42

0.58

1.42

0.58

9.085

9.088

9.085

10.370

7.

0.58

1.42

10.45

10.186

10.450

11.723

8.

1

10.495

10.427

10.495

11.006

9.

0.58

1.42

0.58

1.42

12.8

12.435

12.799

14.116

10.

1.42

0.58

1.42

0.58

1.42

11.21

10.103

10.353

11.858

11.

2

1

8.73

8.474

8.729

8.036

12.

1.42

0.58

6.98

7.169

6.980

6.430

13.

1.42

0.58

10.91

10.592

10.909

10.578

14.

0.58

1.42

15.14

14.811

15.140

14.097

15.

1.42

0.58

1.42

9.58

10.145

9.435

8.883

16.

0.58

1.42

16.2048

16.285

16.204

16.043

17.

1

0

1

7.795

8.070

7.795

8.325

18.

1

0

1

7.81

7.660

7.809

7.702

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

17

Volume 2, 2023

Table 2: Comparison of experimental output and outputs

predicted by Artificial neural network models trained with different algorithms.

4. Conclusion

Neural computing presents a systematic and

pragmatic methodology for representing the

intricate patterns of growth and development in

biological systems, both within and outside the

context of in vitro experimentation. This approach

offers a common endeavour that requires minimal

time and utilizes the available information to its

fullest extent [20]. The approach described in this

study possesses several notable advantages. One

such advantage is its independence from any prior

knowledge about the structure or interrelationships

between input and output signals [21]. The

Levenberg-Marquardt algorithm has been identified

as the most appropriate training algorithm for

predicting the number of shoots. In the context of

19.

0.58

1.42

0.58

1.42

11.42

11.363

11.419

10.670

20.

1

0

10.17

9.866

11.669

9.655

21.

0

1

14.48

14.305

14.480

13.811

22.

1.42

0.58

1.42

8.31

8.723

8.3100

9.3197

23.

0.58

1.42

0.58

11.68

12.0249

11.679

13.550

24.

0.58

1.42

0.58

11.68

11.344

11.679

11.643

25.

1

2

1

9.34

9.735

9.339

9.70

26.

1.42

0.58

7.41

7.0235

7.409

6.448

27.

1.42

0.58

1.42

0.58

8.64

8.097

8.639

8.0375

28.

1

2

12.83

12.364

12.830

12.042

29.

0.58

1.42

0.58

1.42

9.25

9.181

9.249

9.664

30.

0.58

1.42

0.58

14.59

13.671

14.590

13.782

31.

0.58

1.42

0.58

9.01

8.9481

8.995

8.317

32.

1.42

0.58

1.42

0.58

8.44

8.586

8.440

9.224

33.

0.58

1.42

0.58

1.42

0.58

10.52

10.558

10.519

10.768

34.

1.42

0.58

1.42

0.58

12.31

11.678

12.309

12.505

35.

0.58

10.21

9.771

10.209

9.514

36.

1.42

0.58

1.42

10.11

10.083

10.110

9.251

37.

1.42

0.58

1.42

8.84

8.755

8.839

8.761

38.

1

2

1

12.36

13.288

11.823

12.874

39.

1

0

1

11.73

11.227

12.229

13.134

40.

0.58

1.42

0.58

14.96

14.886

14.960

14.731

41.

0.58

1.42

13.34

12.905

13.340

12.971

42.

1.42

0.58

8.36

8.768

8.3600

8.857

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

18

Volume 2, 2023

artificial neural networks, the outcomes derived

from these networks can serve as a fitness function

for optimizing the process. The findings of this study

have significant implications for the application of

optimization algorithms in predicting the optimal

macronutrient composition for achieving maximum

shoot biomass.

The precise estimation of macronutrient levels has

the potential to significantly reduce the financial

burden associated with the formulation of nutritional

media. The efficiency of prediction ability of

Artificial Neural Networks (ANN) is heavily

dependent on the selection of an appropriate training

algorithm. The identification of the optimal training

algorithm plays a crucial role in enhancing the

training and prediction accuracy of plant in vitro

physiological parameters. By incorporating suitable

inputs into the algorithm, the prediction ability of

ANN can be significantly improved. The utilisation

of artificial neural networks (ANN) in predicting the

optimal macronutrient levels for the growth of

chlorophytum borivilianum has proven to be

beneficial for both industrialists and researchers.

This approach offers the potential to cultivate this

plant species at a considerably reduced cost.

References

[1]. Khanam, Zakia, Ompal Singh, Rampal Singh,

and Irshad Ul Haq Bhat. "Safed Musli

(Chlorophytum borivilianum): A review of its

botany, ethnopharmacology and phytochemistry."

Journal of Ethnopharmacology 150, no. 2 (2013):

421-441.

[2]. Thakur, Gulab S., Manoranjan Bag, Bhagwan

S. Sanodiya, Mousumi Debnath, Anish Zacharia,

Pratiksha Bhadauriya, G. B. K. S. Prasad, and P. S.

Bisen. "Chlorophytum borivilianum: a white gold

for biopharmaceuticals and neutraceuticals."

Current Pharmaceutical Biotechnology 10, no. 7

(2009): 650-666.

[3]. Li, Jing, Ji-hang Cheng, Jing-yuan Shi, and Fei

Huang. "Brief introduction of backpropagation

(BP) neural network algorithm and its

improvement." In Advances in Computer Science

and Information Engineering: Volume 2, pp. 553-

558. Springer Berlin Heidelberg, 2012.

[4]. Buscema, Massimo. "Backpropagation neural

networks." Substance use & misuse 33, no. 2

(1998): 233-270.

[5]. Baghirli, Orkhan. "Comparison of Lavenberg-

Marquardt, scaled conjugate gradient and Bayesian

regularization backpropagation algorithms for

multistep ahead wind speed forecasting using

multilayer perceptron feedforward neural network."

(2015).

[6]. Chinatamby, Pavithra, and Jegalakshimi

Jewaratnam. "A performance comparison study on

prediction at industrial areas using different training

algorithms of feedforwardbackpropagation neural

network (FBNN)." Chemosphere 317 (2023):

137788.

[7]. H. Wu, Y. Zhou, Q. Luo, and M. A. Basset,

“Training feedforward neural networks using

symbiotic organisms search algorithm,”

Computational Intelligence and Neuroscience, vol.

2016, Article ID 9063065, 14 pages, 2016.

[8]. F. Burden and D. Winkler, “Bayesian

regularization of neural networks,” Methods in

Molecular Biology, vol. 458, pp. 23–42, 2008.

[9]. L. M. Saini and M. K. Soni, “Artificial neural

network based peak load forecasting using

Levenberg-Marquardt and Quasi-Newton

methods,” IEE Proceedings-Generation,

Transmission and Distribution, vol. 149, no. 5, pp.

578–584, 2002.

[10]. Aich, Ankit, Amit Dutta, and Aruna

Chakraborty. "A scaled conjugate gradient

backpropagation algorithm for keyword

extraction." In Information Systems Design and

Intelligent Applications: Proceedings of Fourth

International Conference INDIA 2017, pp. 674-

684. Springer Singapore, 2018.

[11]. Yu, Hao, and Bogdan M. Wilamowski.

"Levenberg–marquardt training." In Intelligent

systems, pp. 12-1. CRC Press, 2018.

[12]. Sorich, Michael J., John O. Miners, Ross A.

McKinnon, David A. Winkler, Frank R. Burden,

and Paul A. Smith. "Comparison of linear and

nonlinear classification algorithms for the

prediction of drug and chemical metabolism by

human UDPglucuronosyltransferase isoforms."

Journal of chemical information and computer

sciences 43, no. 6 (2003): 2019-2024.

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

19

Volume 2, 2023

[13]. Babani, Lochan, Sadhana Jadhav, and

Bhalchandra Chaudhari. "Scaled conjugate gradient

based adaptive ANN control for SVMDTC

induction motor drive." In Artificial Intelligence

Applications and Innovations: 12th IFIP WG 12.5

International Conference and Workshops, AIAI

2016, Thessaloniki, Greece, September 16-18,

2016, Proceedings 12, pp. 384-395. Springer

International Publishing, 2016.

[14]. Zehra, Andleeb, Mukesh Meena, Dhanaji M.

Jadhav, Prashant Swapnil, and Harish. "Regulatory

Mechanisms for the Conservation of Endangered

Plant Species, Chlorophytum tuberosum—Potential

Medicinal Plant Species." Sustainability 15, no. 8

(2023): 6406.

[15]. Malabadi, Ravindra B., T. L. Nethravathi,

Kiran P. Kolkar, Raju K. Chalannavar, B. S.

Mudigoudra, L. Lavanya, G. Abdi, and H. Baijnath.

"Cannabis sativa: Applications of Artificial

intelligence (AI) and plant tissue culture for

micropropagation." International Journal of

Research and Innovations in Applied Science

(IJRIAS) 8, no. 6 (2023): 117-142.

[16]. Dufera, Abdisa G., Tiantian Liu, and Jin Xu.

"Regression models of Pearson correlation

coefficient." Statistical Theory and Related Fields

(2023): 1-10.

[17]. Torabi, Mahmoud, and J. N. K. Rao.

"Estimation of mean squared error of modelbased

estimators of small area means under a nested error

linear regression model." Journal of Multivariate

Analysis 117 (2013): 76-87

[18]. Rojas Montaño, Razer Anthom Nizer, Carlos

Roberto Sanquetta, Jaime Wojciechowski, Eduardo

Mattar, Ana Paula Dalla Corte, and Eduardo Todt.

"Artificial intelligence models to estimate biomass

of tropical forest trees." Polibits 56 (2017): 29-37.

[19]. Honda H, Takikawa N, Noguchi H, Hanai T,

Kobayashi T (1997) Image analysis associated with

fuzzy neural network and estimation of shoot length

of regenerated rice callus. J Ferment Bioeng

84:342–347. doi:10.1016/S0922-338X(97)89256-2

[20]. Mijwel, Maad M. "Artificial neural networks

advantages and disadvantages." Retrieved from

LinkedIn https//www. linkedin.

com/pulse/artificial-neuralnet Work (2018): 21.

[21]. Prasad, Archana, Om Prakash, Shakti

Mehrotra, Feroz Khan, Ajay Kumar Mathur, and

Archana Mathur. "Artificial neural networkbased

model for the prediction of optimal growth and

culture conditions for maximum biomass

accumulation in multiple shoot cultures of Centella

asiatica." Protoplasma 254 (2017): 335-341.

[22]. Dhanda, Poonam, Subhash Kajla, and

Priyanka Siwach. "Effect of various carbon sources

and gelling agents on in vitro shoot multiplication

of Chlorophytum borivilianum: A medicinal plant

with golden roots." (2023).

[23]. Heng, Seah Yi, et al. "Artificial neural

network model with different backpropagation

algorithms and meteorological data for solar

radiation prediction." Scientific reports 12.1 (2022):

10457.

[24]. Haring, Mark, et al. "A Levenberg- Marquardt

algorithm for sparse identification of dynamical

systems." IEEE Transactions on Neural Networks

and Learning Systems (2022).Tian, Ye, et al.

"Integrating conjugate gradients into evolutionary

algorithms for largescale continuous multi-

objective optimization." IEEE/CAA journal of

automatica sinica 9.10 (2022): 1801-1817.

[25]. Fiorentini, Nicholas, Diletta Pellegrini, and

Massimo Losa. "Overfitting prevention in accident

prediction models: Bayesian regularization of

artificial neural networks." Transportation research

record 2677.2 (2023): 1455-1470.

Contribution of Individual Authors to the

Creation of a Scientific Article (Ghostwriting

Policy)

The authors equally contributed in the present

research, at all stages from the formulation of the

problem to the final findings and solution.

Sources of Funding for Research Presented in a

Scientific Article or Scientific Article Itself

No funding was received for conducting this study.

Conflict of Interest

The authors have no conflicts of interest to declare

that are relevant to the content of this article.

Creative Commons Attribution License 4.0

(Attribution 4.0 International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en

_US

International Journal of Applied Sciences & Development

DOI: 10.37394/232029.2023.2.2

Preeti Kaushik, Neha Khurana,

Madhu Rani, Gopal Krishan, Sonia Kapoor

E-ISSN: 2945-0454

20

Volume 2, 2023