You are on page 1of 19

Contents

1.0 Introduction...........................................................................................................................................3
1.1 Overview of Machine Intelligence and Its Applications.....................................................................3
1.2. Importance of Exploring Specific Areas within Machine Intelligence...............................................3
1.3. Thesis Statement: The Use of Machine Learning Algorithms in Fruit Classification..........................3
1.4 Background........................................................................................................................................3
1.4.1. Definition of Fruit Classification.................................................................................................3
1.4.2. Significance of Accurate Fruit Classification...............................................................................4
1.4.3. Historical Context of Traditional Methods vs. Machine Learning in Classification.....................4
2.0. Literature Review.................................................................................................................................4
2.1 Critique of the Advantages and Disadvantages of Machine Learning in Fruit Classification..............5
3.0 Methodology........................................................................................................................................6
3.1. Explanation of the Chosen Machine Learning Algorithm for Fruit Classification..............................6
3.2. Data Collection Methods and Sources..............................................................................................6
3.3. Preprocessing Steps to Enhance Classification Accuracy..................................................................7
4.0. Critical Analysis.....................................................................................................................................8
4.1. Evaluation of the Chosen Algorithm's Effectiveness.............................................................................8
4.2 . Comparison with Traditional Methods............................................................................................9
4.3. Challenges and Limitations in Fruit Classification Using Machine Learning......................................9
5.0 Future Directions.................................................................................................................................10
5.1 Potential Advancements in Machine Learning for Fruit Classification.............................................10
5.2. Emerging Technologies that May Impact the Field.........................................................................10
5.3 . Recommendations for Further Research and Development..........................................................11
6.0 . Conclusion.........................................................................................................................................12
6.1 . Summary of Key Findings...............................................................................................................12
6.2 . Reiteration of the Significance of Machine Learning in Fruit Classification...................................12
6.3 . Closing Thoughts on the Future of This Research Area..................................................................13
References.................................................................................................................................................13
VIII. Appendices.........................................................................................................................................14
Appendix 1: Training Accuracy Comparison of Machine Learning Algorithms..........................................14
Appendix 2: Distribution of Fruit Classes in Training Dataset................................................................15
Appendix 3: Code Snippets for Preprocessing Steps.............................................................................21
Appendix 4: Details of Machine Learning Model Architecture.............................................................21
FRUIT CLASSIFICATION USING MACHINE LEARNING ALGORITHMS
1.0 Introduction machine intelligence in the study
1.1 Overview of Machine Intelligence and environment.
Its Applications
1.3. Thesis Statement: The Use of Machine
A sub-area of artificial intelligence involves Learning Algorithms in Fruit Classification
various learning techniques and algorithms, The topic of this article entails the
which aim at making the machines self- specificity of one application in Machine
learning and decision-making, depending on Intelligence, fruit classification, and it
data that is being fed into them without involves reviewing the usage of the machine
specific instructions (Abdullayeva and learning algorithms. This study will focus at
Engalichev, 2022 and Xash Machine the interface of technology and agriculture
intelligence is utilized in various sectors like where machine learning will be tested for its
health, finances, transport logistics, etc. accuracy of fruit classification based on
Machine learning algorithms have been at different characteristics. This essay
the heart of rapid advances that are evaluates the possible benefits, limitations
transforming this fast-evolving area. and probable threats of using machine
learning in fruit classification by critical
1.2. Importance of Exploring Specific Areas
within Machine Intelligence examination of literature.

Although Machine intelligence is broad and 1.4 Background


dynamic, we need to focus on some of the
1.4.1. Definition of Fruit Classification
points to gain more insight and discover the The identification of fruits is done under
uses in different sectors. Specialized search fruit classification; whereby there are
means researches and developments with categories formed that focus on factors like
respect to narrow scopes. By focusing on size, color, shape, and even texture. The
these issues we gain better understanding reason is that this procedure involves
into the essence of those concepts while at systematic order of fruits which makes it
the same time we are able to come with simple for farmers, deliverers, and
feasible solutions to modern day challenges. customers to specify and administer various
This intensive exploration allows the student types of fruits. That is why lately automated
to interact with one specific element of the techniques for fruit classification, which
exploit technology especially the one based makes use of big data sets, which are used
on machine learning, are being preferred for developing algorithms that can
over traditional human expertise. accurately identify fruit classification. The
move from conventional to automated
techniques indicate a model of change in
1.4.2. Significance of Accurate Fruit undertaking fruit categorization that has
Classification
Fruit classification is very critical for several better chances of being efficient and correct.
fields especially in agriculture and food
marketing. Accurate categorization helps in 2.0. Literature Review
crop management for farmers as they can Automation of fruit categorization

plan to maximize harvesting, storage and operations and enhancement of reliability

distribution aspects. Accurate fruit and effectiveness can only be possible

classification, which entails sorting, through machine learning techniques. A

packaging, and delivery of products without number of techniques were deployed with an

waste in a supply chain, promotes product aim of getting the best results in this area.

effectiveness. Moreover, good SVM is well known for its capability of

categorization of fruit for consumers assists tackling complicated classification issues by

them in making informed purchases as well. classifying fruits into different categories
using variables like color texture and shape.
1.4.3. Historical Context of Traditional
The use of random forest is helpful
Methods vs. Machine Learning in
Classification especially dealing with big datasets,
Traditionally, the identification of different showing tolerance to over fitting and
fruits has been done through a visual achieving high precision during fruits
inspection by experts and grouping together classification job. In addition, studies have
similar features. This approach may be shown promise with CNN-based classifiers
successful but consumes considerable for fruit classification using visual cues.
resources such as manpower and time Lastly, K-Nearest Neighbours (KNN)
besides being subjective due to individual algorithms which depend on proximity also
competence of a classifier. The advancement employ this approach of predicting about
of machine learning, and more so over the how close a piece of fruit is to its neighbours
last two decades, has given rise to automated in the feature space.
classification systems. Such type of system
Various machine learning techniques 2.1 Critique of the Advantages and
Disadvantages of Machine Learning in
facilitate automatic fruit classification and
Fruit Classification
offer diverse solutions to many tasks within The use of machine learning in fruit
this classification such as quick and accurate classification facilitates increased efficiency
processing. and greater precision in the whole exercise.
There is one important benefit, namely,
The application of machine learning
enhanced speed and automation that cut
techniques has turned out to be effective in
down on the time and the efforts associated
numerous cases of fruit classification.
with manual sorting. Similarly, trained
SVM’s are among the algorithms that assist
machine learning algorithms reach
in objective and standardized assessment of
surprising levels of correctness, often
fruit quality in automated fruit grading.
exceeding the accuracy of human experts in
These systems grade parameters like size or
discriminating fruits(Chaudhari and
color or nature of flaw in order to ensure
Waghmare, 2022; Naik and Patel, 2017)..
proper grading (Ramya et al., 2020). These
Also, these models are flexible as they are
include, use of picture analysis algorithm
able to respond to different datasets and
like Random forest to determine fruit
changing conditions while classifying
maturity, planning of harvesting seasons so
different fruit types under different
that supply of the highest quality and fresh
conditions.
fruits can be achieved to the consumer.4
Another innovative application is multi- Although, there are some problems with
sensor fusion for classification that machine learning in fruit classification. This
combines information from various sensors is one of the most critical aspects since it
like colour, texture, weight data, thereby relies on high-quality and representative
increasing classification accuracy for training data. This can be traced by how
machine learning based fruit related systems machine learning models depend on the
as a testimony to how far reaching machine training datasets. The use of defective or
learning is. biased data in training the machine learning
model can cause misleading predictions
(Bhatt et al., 2023). A crucial point in this
regard is the fact that complex models, for
example deep neural networks, can also turn
out to be hardly interpretable and transparent The functionality of CNN to work on
when making decisions. The lack of different levels of abstractness from which it
interpretability could impede gradually extracts highly complicated traits
comprehension, especially about why is consistent with its aptness for
certain categories are created. Additionally, complexities present in image data about
training and implementing sophisticated fruits.
machine learning algorithms may require
As such, using the power of CNNs is
extensive computing power that is not
consistent with this aim of improving the
readily available in impoverished settings.
degree of classification accuracy of fruits
through visual features. These are attributes
3.0 Methodology that fall on a continuum incorporating color,
3.1. Explanation of the Chosen Machine texture, and shape, but which give distinct
Learning Algorithm for Fruit Classification
Therefore, this research has used the CNN’s look to various kinds of fruit. This research

machine learning model as a unique option expects that by having a CNN classify these

for the complex classification job of fruits. visual properties a more comprehensive and

This is a deliberate decision based on the detailed knowledge concerning fruit

fact that CNN’s have been quite good for classification will be achieved thus creating

image based tasks over and over again. They a more accurate fruit classification model.

have this intrinsic ability to identify and Among all machine learning algorithms,

interpret the various visual attributes that are CNN has an inherent ability for hierarchical

inherently linked with fruit classification feature extraction and is poised to unlock

because they are well endowed. mysteries hidden in the images from the
visual world that captures fruits.
A CNN is presented as a complex textile of
connected layers that focus on convolution 3.2. Data Collection Methods and Sources
The success of a machine learning model
layers. The layers, just like an attentive
heavily relies on the quality and diversity of
painter’s strokes, help learn hierachrical
the training data. For this study, a
features from input images. The hierarchical
comprehensive dataset of fruit images will
learning approach underpins the search of
be collected from various sources. These
complex structures which allow
sources may include:
classification of thousands of fruit varieties.
1. Public Databases: Utilizing publicly that empowers the machine learning model
available datasets, such as the Fruits-360 for optimal performance.
dataset, allows for a broad representation of 1. Image Resizing and Normalization:
different fruit classes. These datasets often The initial step involves the
come with labeled images, streamlining the standardization of input images
training process. through resizing. This not only
2. Field Image Capture: Capturing images ensures a consistent format but also
directly from fruit orchards ensures a mitigates potential distortions,
realistic representation of fruits in their fostering uniformity during the
natural environment. This approach helps training process. Simultaneously,
the model generalize well to real-world normalizing pixel values is
scenarios. undertaken to alleviate the impact of
variations in illumination. This dual-
3. Collaboration with Agricultural
pronged approach sets the foundation
Institutions: Collaborating with agricultural
for a model capable of robustly
research institutions can provide access to
handling diverse image inputs.
specialized datasets, ensuring a diverse
representation of fruits and accounting for 2. Data Augmentation: A pivotal facet
variations in growing conditions. in the preprocessing journey is data
augmentation. Techniques
3.3. Preprocessing Steps to Enhance
Classification Accuracy encompassing rotation, flipping, and
In the pursuit of refining the accuracy of the zooming are harnessed to diversify
fruit classification model, a meticulously the training dataset. By introducing
crafted set of preprocessing steps is these variations, the risk of
envisaged. Each step in this methodological overfitting diminishes, and the
approach is meticulously designed to model's generalization capabilities
augment the model's capabilities, ensuring it are fortified. This strategic
can discern and classify fruits with a augmentation aims to imbue the
heightened level of precision. The model with adaptability, enabling it
overarching goal is not merely to process to discern fruits under varying
data but to sculpt a preparatory framework conditions.
3. Feature Extraction: The model's ability to navigate the complexities
incorporation of pre-trained models, of fruit classification based on visual
such as VGG16 or ResNet, embodies attributes. This methodological finesse
a sophisticated layer in the doesn't merely aim for incremental
preprocessing pipeline. This improvements; it aspires to cultivate a
technique, known as transfer machine learning model that epitomizes
learning, allows the model to glean robustness and efficacy in the intricate realm
high-level features from images. of fruit classification.
Particularly advantageous when
faced with limited labeled data, 4.0. Critical Analysis
feature extraction propels the model
beyond basic image analysis, 4.1. Evaluation of the Chosen Algorithm's
Effectiveness
enriching its discernment of intricate
The Convolutional Neural Network (CNN)
patterns.
selected as the linchpin for fruit
4. Label Encoding: A pivotal bridge classification undergoes a meticulous
between categorical labels and evaluation, unraveling its effectiveness
numerical compatibility is across multiple dimensions. The chosen
constructed through label encoding. model demonstrates prowess in not only
This step ensures harmonious capturing but deciphering intricate patterns
interaction between the machine and features pivotal for the nuanced
learning model and the training data. differentiation of diverse fruit classes based
By converting categorical labels into on their visual attributes. This efficacy is not
a numerical format, the model can merely superficial; it delves deep into the
seamlessly ingest and process the training process, where the CNN, as an
diverse information encapsulated autonomous learner, assimilates hierarchical
within the dataset. representations. These learned

The orchestration of these preprocessing representations empower the model to make

steps is akin to fine-tuning an instrument nuanced decisions, reflecting a sophisticated

before a performance. Each nuanced understanding of the visual intricacies

adjustment serves a specific purpose, inherent in various fruit varieties.

collectively harmonizing to enhance the


The forthcoming assessment of the practical deployment in real-world
algorithm's effectiveness encompasses a applications.
comprehensive suite of performance
4.2 . Comparison with Traditional Methods
metrics. These metrics extend beyond the Comparing the chosen CNN-based approach
conventional accuracy measurement, with traditional methods, such as manual
embracing precision, recall, and the F1 classification by human experts, reveals
score. This multifaceted evaluation aims to several advantages and trade-offs. The CNN
provide a holistic understanding of the exhibits efficiency by automating the
algorithm's classification capabilities, classification process, significantly reducing
transcending a one-dimensional assessment. the time and labor required for manual
Moreover, the scrutiny of the chosen inspection. Moreover, the algorithm has the
algorithm doesn't conclude with its potential to outperform traditional methods
performance on the training dataset. The in terms of accuracy, especially when
model's true mettle is revealed through its dealing with large datasets and complex
ability to generalize to unseen data—a patterns. However, the interpretability of the
litmus test for its adaptability to real-world CNN may pose challenges, as understanding
scenarios. This extends to an exploration of the decision-making process within the
how well the algorithm contends with neural network is less straightforward
variations in fruit appearance, an aspect compared to the transparent nature of human
critical for practical applicability in diverse expertise. This analysis aims to highlight the
agricultural settings. strengths and limitations of both approaches,
providing a comprehensive view of the
As the critical analysis unfolds, it ventures
applicability of machine learning in fruit
beyond a cursory examination of accuracy
classification.
and delves into the nuanced intricacies of
the algorithm's decision-making process. 4.3. Challenges and Limitations in Fruit
Classification Using Machine Learning
This depth of analysis ensures that the
Despite the advancements in machine
evaluation isn't confined to isolated metrics
learning for fruit classification, several
but embraces a comprehensive
challenges and limitations persist:
understanding of the algorithm's
1. Data Quality and Quantity: The
functionality, robustness, and potential for
effectiveness of machine learning models
heavily relies on the quality and quantity of 5.0 Future Directions
training data. Inadequate or biased datasets 5.1 Potential Advancements in Machine
can lead to suboptimal performance and Learning for Fruit Classification
Advanced Neural Network Architectures:
misclassifications.
The trajectory of machine learning for fruit
2. Environmental Variability: Fruits can classification foresees the evolution of
exhibit variations in appearance due to neural network architectures. Innovations in
environmental factors, such as lighting this domain may give rise to more
conditions and background clutter. Adapting sophisticated Convolutional Neural Network
machine learning models to handle these (CNN) variants or attention mechanisms.
variations poses a significant challenge. These advancements hold the promise of
3. Generalization to Unseen Varieties: empowering machine learning models to
Machine learning models may struggle to capture intricate features and subtleties
generalize effectively to unseen fruit present in fruit images, thereby leading to a
varieties not present in the training dataset. substantial improvement in classification
This limitation necessitates ongoing efforts accuracy.
to expand and diversify training data. i. Transfer Learning Techniques: The
4. Interpretability: The lack of future landscape of fruit
interpretability in some complex machine classification is poised for further
learning models, including CNNs, can exploration of transfer learning
hinder the understanding of how specific methodologies. Leveraging pre-
decisions are made. This lack of trained models on expansive datasets
transparency may raise concerns in practical and fine-tuning them specifically for
applications where interpretability is crucial. fruit classification tasks can
streamline the development process.
5. Computational Resources: Training and
This approach becomes particularly
deploying sophisticated machine learning
invaluable when confronted with
models, especially deep neural networks,
challenges related to limited labeled
require substantial computational resources.
data, as the model can capitalize on
This can limit the accessibility of advanced
the knowledge acquired from diverse
models in resource-constrained
datasets.
environments.
ii. Ensemble Learning Approaches: The environmental variations on model
horizon of machine learning performance.
applications in fruit classification ii. Explainable AI (XAI): As machine
includes the investigation of learning models become increasingly
ensemble learning techniques. By sophisticated, the integration of
delving into the combination of explainable AI techniques gains
predictions from multiple machine prominence. This technological
learning models, this approach aims avenue addresses the interpretability
to fortify the robustness and challenge associated with complex
generalization capabilities of the models. Ensuring transparency in
models. Ensemble learning holds decision-making processes is crucial,
significant potential in addressing the especially in applications where
complexities arising from variations human understanding is
in fruit appearance across diverse indispensable for fostering trust in
contexts. the model's outputs.
5.2. Emerging Technologies that May iii. Edge Computing: The paradigm of
Impact the Field
i. Sensor Fusion: The integration of deploying machine learning models

data from an array of sensors, on edge devices, closer to the point

encompassing RGB cameras, of data generation, stands as a

hyperspectral imaging, and other transformative approach. Particularly

advanced imaging technologies, relevant in agriculture, where real-

emerges as a pivotal technological time processing is paramount, edge

frontier. Sensor fusion has the computing enhances efficiency by

potential to furnish a more holistic reducing latency. This decentralized

and nuanced understanding of fruit approach also diminishes the reliance

characteristics. This comprehensive on centralized computational

insight is instrumental in enhancing resources, making it well-suited for

classification accuracy and resource-constrained environments.

mitigating the impact of


5.3 . Recommendations for Further cornerstone for further advancements.
Research and Development
Collaborative and interdisciplinary
Diverse and Representative Datasets: The
approaches are instrumental in developing
trajectory of fruitful research in machine
holistic solutions. Such collaboration
learning for fruit classification necessitates
ensures that fruit classification models are
persistent efforts in curating diverse and
not only technologically sophisticated but
representative datasets. The richness of these
also align with the nuanced requirements
datasets is paramount for training models
and perspectives of the agricultural domain.
that generalize effectively across various
fruit varieties, growing conditions, and By delving into these potential
environmental factors. advancements, embracing emerging
technologies, and committing to further
Addressing Bias and Fairness: A crucial
research and development, the field of
avenue for further research involves
machine learning for fruit classification is
developing methodologies to identify and
poised to transcend existing limitations,
mitigate biases within machine learning
paving the way for more robust, adaptable,
models. This commitment is essential for
and practical applications in agriculture and
ensuring fair and equitable outcomes in fruit
beyond.
classification across different demographic
and environmental contexts.
6.0 . Conclusion
User Feedback Integration: The adaptability
6.1 . Summary of Key Findings
of fruit classification systems can be
In the culmination of this comprehensive
significantly enhanced through the
exploration into the application of machine
incorporation of user feedback into the
learning for fruit classification, with a
machine learning pipeline. Continuous
specific focus on Convolutional Neural
learning mechanisms that take user input
Networks (CNNs), the key findings
into account foster iterative improvements,
encapsulate a multifaceted evaluation. This
resulting in enhanced model performance
investigation delved into the efficacy of the
over time.
chosen CNN algorithm, conducted a
Interdisciplinary Collaboration: The synergy comparative analysis vis-à-vis traditional
between computer scientists, agricultural methods, and critically dissected the
experts, and data scientists emerges as a
challenges and limitations inherent in fruit with the ongoing technological
classification using machine learning. metamorphosis in agriculture, the
integration of machine learning signals a
The CNN, as the focal point of this study,
trajectory towards more data-driven and
exhibited notable efficacy in deciphering
automated approaches.
intricate patterns essential for distinguishing
between diverse fruit classes. Through 6.3 . Closing Thoughts on the Future of
comparative analysis, the study shed light on This Research Area
The contemplation of the future trajectory of
the efficiency gains realized by automated
machine learning in fruit classification
classification in contrast to conventional
unveils a landscape brimming with promise
manual methods. However, a discerning
and potential for further exploration.
examination acknowledged challenges
Ongoing advancements in neural network
pertaining to interpretability and data
architectures, the nuanced application of
dependency, adding nuance to the
transfer learning techniques, and the
understanding of machine learning's role in
seamless integration of emerging
fruit classification.
technologies such as sensor fusion and
6.2 . Reiteration of the Significance of explainable AI collectively stand as pillars
Machine Learning in Fruit Classification
poised to elevate the accuracy and
The enduring significance of machine
applicability of fruit classification models.
learning in fruit classification reverberates
throughout agricultural practices and supply Crucial to this future narrative is the
chain management. The adoption of interdisciplinary collaboration that
automated fruit classification systems, converges computer scientists, agricultural
empowered by machine learning algorithms, experts, and data scientists. This
presents a paradigm shift characterized by collaborative synergy is not merely
heightened efficiency, accuracy, and advantageous but rather imperative for
scalability. The expeditious and accurate addressing the intricacies and challenges
categorization of fruits based on visual inherent in the agricultural domain. The
attributes not only streamlines operations for future unfolds with a commitment to
farmers and distributors but also augments integrating user feedback, fostering
the overall quality and consistency of continuous learning mechanisms, and
products reaching consumers. In alignment navigating the ethical considerations
associated with biases, contributing to the Singapore.
responsible and equitable deployment of https://link.springer.com/chapter/10.
machine learning models in agriculture. 1007/978-981-16-7985-8_81
3. GILL, H.S., MURUGESAN, G.,
In anticipation, the dynamic evolution of
MEHBODNIYA, A., SAJJA, G.S.,
technologies and methodologies continues to
GUPTA, G. AND BHATT, A.,
shape the future of fruit classification using
2023. Fruit type classification using
machine learning. This field stands
deep learning and feature
resiliently at the forefront of innovation,
fusion. Computers and Electronics in
destined to leave an indelible mark on the
Agriculture, 211, p.107990.
agricultural landscape, forging a path
https://www.sciencedirect.com/scien
towards sustainable, efficient, and
ce/article/pii/S0168169923003782
technologically empowered practices.
4. NAIK, S. AND PATEL, B., 2017.
Machine vision based fruit
References
1. ABDULLAYEVA, O., & classification and grading-a
ENGALICHEV, M., 2022. Artificial review. International Journal of
intelligence systems. Значение Computer Applications, 170(9),
Цифровых Технологий В Изучении pp.22-34.
Истории Узбекистана, 1(01), 382– https://www.researchgate.net/profile/
385. Sapan-Naik/publication/318486455_
https://doi.org/10.47689/.v1i01.1361 Machine_Vision_based_Fruit_Classi
2 fication_and_Grading_-_A_Review/
2. CHAUDHARI, D. AND links/5ca1b731a6fdccd46047f7da/
WAGHMARE, S., 2022, May. Machine-Vision-based-Fruit-
Machine vision based fruit Classification-and-Grading-A-
classification and grading—a review. Review.pdf
In ICCCE 2021: Proceedings of the 5. SARANYA, N., SRINIVASAN, K.,
4th International Conference on PRAVIN KUMAR, S. K.,
Communications and Cyber Physical RUKKUMANI, V., & RAMYA, R.,
Engineering (pp. 775-781). 2020. Fruit classification using
Singapore: Springer Nature traditional machine learning and
deep learning approach.
In Computational Vision and Bio-
Inspired Computing: ICCVBIC
2019 (pp. 79-89). Springer
International Publishing. Fruit
Classification Using Traditional
Machine Learning and Deep
Learning Approach | SpringerLink
6. XASHIMOV, B., &
KHAYDAROVA, D., 2023. Using This chart visually represents the
and development of artificial comparative training accuracy of different
intelligence on the process of machine learning algorithms in fruit
accounting. Новый Узбекистан: classification. The data is based on
Успешный Международный Опыт experimental results obtained during the
Внедрения Международных evaluation phase of the study.
Стандартов Финансовой
Appendix 2: Distribution of Fruit Classes
Отчетности, 1(5), 219–223. in Training Dataset
https://doi.org/10.47689/stars.univers
ity-5-pp219-223

VIII. Appendices

Appendix 1: Training Accuracy


Comparison of Machine Learning
Algorithms

This graph displays the distribution of


different fruit classes within the training
dataset. Understanding the class distribution
is crucial for assessing the potential biases Apple:
and challenges associated with the dataset.

Banana:

Oranges:
Fresh:

Rotten:

Inference:
B. Additional Information Supporting the
Main Content
Appendix 3: Code Snippets for
Preprocessing Steps

These code snippets provide insight into the


These code snippets provide a glimpse into architecture of the Convolutional Neural
the preprocessing steps applied to the Network used for fruit classification. The
images before feeding them into the model's layers, activation functions, and
Convolutional Neural Network. compilation settings are outlined to facilitate
Understanding these steps is crucial for a deeper understanding of its structure.
replicating and extending the study.

Appendix 4: Details of Machine Learning


Model Architecture

You might also like