Artificial

Intelligence
&
Machine
Learning
 Team Members

BGNE Kumarasinghe YBWWALY Bandara

 Table of contents
01 • Overview of AI and ML
02 • Applications of AL and ML
03 • Research Trends
04 • Future Direction of AI and ML
05 • What do we think about AI and ML?
06 • References
 Overview of AI and ML
Artificial Intelligence

Artificial intelligence (AI) is a broad field of computer science that

aims to create intelligent machines. This can be achieved through
various techniques, including machine learning, natural language
processing, computer vision, and robotics. The goal of AI is to create
machines that can think and act like humans.

Machine Learning

Machine learning (ML) is a subfield of AI that focuses on developing

algorithms that can be learned from data. These algorithms can
then make predictions or decisions on new data. Machine learning
algorithms are trained on large amounts of data, and they can
improve their performance over time as they are exposed to more
data.
 Evolution of AI and ML

Early Seeds of AI (1940s-1950s)

• Alan Turing: Besides the Turing Test, Turing also contributed to the concept of the Universal Turing Machine, a theoretical
model that could simulate any computer system.
• John McCarthy: Along with coining "artificial intelligence," McCarthy also held the Dartmouth workshop in 1956, a
landmark event that brought together leading researchers and is considered the official birth of AI research.
• Early approaches: Symbolic logic involved representing knowledge as symbols and rules, while expert systems aimed to
capture the expertise of human specialists in a specific domain. An example could be a medical diagnosis system.

Birth of Machine Learning (1950s-1960s)

• from its experiences, demonstrating the potential of this approach.
• Frank Rosenblatt: His Perceptron, an artificial neural network inspired by the human brain, laid the foundation for future
neural network research.
• Marvin Minsky: While Minsky's work on the limitations of Perceptron contributed to the "AI Winter," his later research on
connectionism (another form of neural networks) remained influential.

Early Challenges and AI Winter (1960s-1980s)

• Limitations: Early AI systems struggled with the complexity of real-world problems and the lack of processing power to
handle large datasets.
• Shifting focus: Research turned towards more specific and achievable goals in areas like natural language processing
• (e.g., machine translation) and computer vision (e.g., object recognition).
Resurgence of AI and Rise of ML (1980s-2000s)
• Computing power: Advancements in hardware, like faster processors, allow for more complex algorithms and larger
datasets.
• Backpropagation: This algorithmic breakthrough enabled efficient training of multi-layered neural networks, a key driver
of deep learning's success.
• Machine learning techniques: Decision trees offered a clear and interpretable way to make decisions based on data,
while support vector machines excelled at classification tasks.

The Deep Learning Revolution (2000s-Present)

• Massive datasets: The explosion of data generated daily (e.g., from social media, and internet searches) provided the fuel
for deep learning algorithms.
• GPUs: Graphics processing units, originally designed for video games, proved exceptionally efficient for training deep
learning models due to their parallel processing capabilities.
• Deep learning breakthroughs: Image recognition software like convolutional neural networks achieved human-level
performance, while recurrent neural networks revolutionized natural language processing (e.g., machine translation, text
generation).

The Present and Future

• Focus areas: Explainable AI aims to make AI models more transparent in their decision-making processes, while robust AI
strives for systems that function reliably in real-world situations.
• Frontiers: Artificial general intelligence (AGI) aspires to create machines with human-level intelligence across all domains
and embodied AI seeks to integrate AI with robots that can interact with the physical world.
Programming vs Machine Learning

Traditional Programming
Rules + Data -> Desired Output

Machine Learning
Rules + Data -> Desired Output
 Types of Deep learning Models and Their Uses

Generative
Convolutional Recurrent Reinforcement
Adversarial
Neural Networks Neural Networks Learning
Networks

Computer vision, Process Sequences Well-known in Generating new

Image classifications ex: Text, Speech, Games, used in artifacts ex: Potos,
Audio Robotics Videos, Music
 Types of Machine Learning

Machine Learning Types

Supervised Unsupervised Semi Reinforcement
Learning Learning Supervised Learning
Learning
Classification Clustering Positive RL
Regression Association Negative RL
 1. Supervised Learning

• Supervised machine learning is based on supervision.

• It utilizes a labeled dataset for training.
• The labeled data pairs inputs with corresponding outputs.
• During training, the machine learns the relationship between inputs and outputs.
• After training, the machine can predict outputs for new inputs.
• Testing is done using a separate dataset not seen during training.
• The test dataset evaluates the machine's predictive performance.
1.1 Classification 1.2 Regression

2. Unsupervised Learning

• Unsupervised learning differs from supervised learning as it doesn't require supervision.

• In unsupervised machine learning, the machine is trained using unlabeled data.
• The machine predicts outputs without any predefined labels or supervision.
• Models in unsupervised learning are trained with unclassified and unlabeled data.
• The main objective is to group or categorize the unsorted dataset based on similarities, patterns, and
differences.
• Machines are tasked with discovering hidden patterns from the input dataset.
2.1 Clustering 2.2 Association
3. Semi-Supervised Learning

• Semi-supervised learning is positioned between Supervised and Unsupervised machine learning.

• It utilizes a combination of labeled and unlabeled datasets during training.
• This approach aims to leverage all available data effectively, unlike supervised learning, which only uses
labeled data.
• Semi-supervised learning addresses the limitations of both supervised and unsupervised learning
algorithms.
• Semi-supervised learning algorithms make use of both labeled and unlabeled data to improve learning
accuracy and generalize better to unseen data.
• This approach is particularly useful when obtaining labeled data is expensive or time-consuming but
unlabeled data is abundant.

4. Reinforcement Learning

• Reinforcement learning operates on a feedback-based process.

• An AI agent explores its environment through trial and error, taking actions and learning from
experiences.
• The agent receives rewards for good actions and penalties for bad ones, aiming to maximize cumulative
rewards.
• Unlike supervised learning, reinforcement learning does not rely on labeled data.
• Agents learn solely from their experiences within the environment.
4.1 Positive RL 4.2 Negative RL
Machine Learning Frameworks
TensorFlow is an open-source framework developed
by Google. It is a powerful and flexible framework that
can be used for a wide variety of machine learning
tasks, including deep learning.

PyTorch is another open-source framework that is

popular for deep learning. PyTorch is known for its
ease of use and flexibility.

Scikit-learn is a free and open-source software library for

machine learning in Python. It features various classification,
regression, clustering, dimensionality reduction, model selection,
and preprocessing algorithms. Scikit-learn is a great choice for
beginners because it is easy to learn and use.

Keras is a high-level neural network API that can be

used on top of TensorFlow, PyTorch, or other deep
learning frameworks. Keras is known for its ease of use
and simplicity.
How Machine Learning Model Train

1. Data Collection
2.Feature Identification
3.Learning & Practice
4.Testing
Tech Giants and the Rise of AI and ML

Google Amazon Netflix

Microsoft Apple Tesla

Advantages, Disadvantages and Limitations of AL and ML

Advantages
·Enhanced Efficiency and Insights
·Personalization
·Improved Accuracy and Automation
·Data-Driven Decision Making Disadvantages
·Advanced Analytics and Discovery ·Job Displacement
·24/7 Availability and Scalability ·Bias and Fairness
·Innovation and Optimization ·Lack of Explainability
·Security and Privacy Risks
·Overdependence on Technology
·Existential Risks

Limitation
• Data Dependence
• Limited Reasoning
• Interpretability
• Creativity and Ethics
• Computational Cost
 Applications of AL and ML
Entertainment
⚬ Creating realistic experiences in games, and movies (e.g., special effects)
⚬ Recommending content based on user preferences

Healthcare
⚬ Diagnosing diseases (image analysis)
⚬ Developing new drugs
⚬ Providing personalized care (predicting health problems)

Finance
⚬ Detecting fraud
⚬ Managing risk
⚬ Making investment decisions (trading algorithms)

Customer Service

⚬ Automating tasks (chatbots)

Manufacturing

⚬ Optimizing factory processes

⚬ Improving product quality
⚬ Predicting maintenance needs

Transportation
⚬ Developing self-driving cars (autonomous vehicles)

Security
⚬ Detecting cyber threats
⚬ Identifying malicious activity

E-commerce
⚬ Personalizing shopping experience
⚬ Recommending products
 AI and ML in Healthcare
Medical Imaging Analysis Drug discovery and development Personalized Medicine

Disease prediction and prevention Robot-assisted surgery Virtual Assistants

Challenges in healthcare using
AI and ML

Data privacy and security

Bias and fairness

Regulation and oversight

ML
 Research Trends

• A. İ. Tekkeşin, “Artificial Intelligence in Healthcare: Past, Present and Future,” Anatolian

01 journal of cardiology, vol. 22. NLM (Medline), pp. 8–9, Oct. 01, 2019. doi:
10.14744/AnatolJCardiol.2019.28661.
• The document "Artificial Intelligence in Healthcare: Past, Present and Future" by
Ahmet İlker Tekkeşin discusses the potential of artificial intelligence (AI) and
machine learning (ML) to revolutionize healthcare by synthesizing vast amounts of
data and experience to enhance patient care. It emphasizes the need for clinicians to
embrace AI and acquire the necessary skills to interpret patient-level data effectively
in the era of digitalized healthcare.[1]
• N. Kühl, M. Goutier, R. Hirt, and G. Satzger, Machine Learning in Artificial

02 Intelligence: Towards a Common Understanding. [Online]. Available:

https://hdl.handle.net/10125/59960
• The document explores the relationship between machine learning and
artificial intelligence, providing a conceptual framework to clarify the role of
machine learning in building intelligent agents. It distinguishes between
simple-reflex and learning agents, highlighting the role of machine learning in
each case and the sublayers of executing and learning backends within
intelligent agents.[2]
 • R. Cioffi, M. Travaglioni, G. Piscitelli, A. Petrillo, and F. De Felice, “Artificial intelligence and
03 machine learning applications in smart production: Progress, trends, and directions,”
Sustainability (Switzerland), vol. 12, no. 2. MDPI, Jan. 01, 2020. doi: 10.3390/su12020492.
• The document presents a comprehensive bibliometric analysis of research on artificial
intelligence (AI) and machine learning (ML) applications in the context of smart
production and sustainability, covering the period from 1999 to 2019. It employs a
systematic literature review methodology, combining bibliometric, content analysis,
and social network techniques to analyze the most influential documents, keywords,
collaborative authors, and the benefits of AI and ML in industrial contexts. [3]

• M. Maadi, H. A. Khorshidi, and U. Aickelin, “A review on human–ai interaction in machine

04 learning and insights for medical applications,” International Journal of Environmental
Research and Public Health, vol. 18, no. 4. MDPI AG, pp. 1–21, Feb. 02, 2021. doi:
10.3390/ijerph18042121.
• The document provides a comprehensive overview of human-AI interaction in
machine learning (ML) applications, focusing on the role of humans in data producing,
ML modeling, and evaluation and refinement stages. It also highlights the importance
of human collaboration with ML methods, particularly in medical applications, and
identifies challenges and future research directions in this area.[4]
 • N. Kühl, M. Schemmer, M. Goutier, and G. Satzger, “Artificial intelligence and machine
05 learning,” Electronic Markets, vol. 32, no. 4, pp. 2235–2244, Dec. 2022, doi: 10.1007/s12525-022-
00598-0.
• The document explores the relationship between artificial intelligence (AI) and
machine learning (ML) in the context of information systems research, aiming to clarify
their usage and provide a conceptual framework for understanding their roles. It
emphasizes the need for terminological precision and guidance in differentiating
between AI and ML, proposing a typology for AI-based information systems to provide
clarity and guidance for future research and interdisciplinary discussions.[5]
• R. Tiwari, “The integration of AI and machine learning in education and its potential to
06 personalize and improve student learning experiences,” INTERANTIONAL JOURNAL OF
SCIENTIFIC RESEARCH IN ENGINEERING AND MANAGEMENT, vol. 07, no. 02, Feb. 2023, doi:
10.55041/ijsrem17645.
• The document is a literature review that examines the integration of artificial
intelligence (AI) and machine learning (ML) in education, focusing on the potential of
AI to personalize and improve student learning experiences, while also addressing the
need for further research to understand the capabilities and limitations of AI in
education and the ethical and societal implications of its use. [6]
 • D. Pattnaik, S. Ray, and R. Raman, “Applications of artificial intelligence and machine
07 learning in the financial services industry: A bibliometric review,” Heliyon, vol. 10, no. 1, Jan.
2024, doi: 10.1016/j.heliyon.2023.e23492.
• The document is a bibliometric review that analyzes the trends and thematic structure
of research on artificial intelligence (AI) and machine learning (ML) applications in the
Banking, Financial Services, and Insurance (BFSI) sector. It employs N-gram and co-
occurrence analyses to identify clusters of research themes, providing insights into the
transformative impact of AI and ML technologies in the BFSI sector and guiding future
scholarly endeavors. [7]
 Future Direction of AI and ML
Increased Focus on Explainable AI (XAI)

Generative AI

AI for Sustainability

Human-AI Collaboration

Ethical Considerations
What Do We Think
about AI and ML?
 References
• [1] A. İ. Tekkeşin, “Artificial Intelligence in Healthcare: Past, Present and Future,” Anatolian
journal of cardiology, vol. 22. NLM (Medline), pp. 8–9, Oct. 01, 2019. doi:
10.14744/AnatolJCardiol.2019.28661.
• [2] N. Kühl, M. Goutier, R. Hirt, and G. Satzger, Machine Learning in Artificial Intelligence: Towards
a Common Understanding. [Online]. Available: https://hdl.handle.net/10125/59960
• [3] R. Cioffi, M. Travaglioni, G. Piscitelli, A. Petrillo, and F. De Felice, “Artificial intelligence and
machine learning applications in smart production: Progress, trends, and directions,” Sustainability
(Switzerland), vol. 12, no. 2. MDPI, Jan. 01, 2020. doi: 10.3390/su12020492.
• [4] M. Maadi, H. A. Khorshidi, and U. Aickelin, “A review on human–ai interaction in machine
learning and insights for medical applications,” International Journal of Environmental Research and
Public Health, vol. 18, no. 4. MDPI AG, pp. 1–21, Feb. 02, 2021. doi: 10.3390/ijerph18042121.
• [5] N. Kühl, M. Schemmer, M. Goutier, and G. Satzger, “Artificial intelligence and machine
learning,” Electronic Markets, vol. 32, no. 4, pp. 2235–2244, Dec. 2022, doi: 10.1007/s12525-022-00598-0.
• [6] R. Tiwari, “The integration of AI and machine learning in education and its potential to
personalize and improve student learning experiences,” INTERANTIONAL JOURNAL OF SCIENTIFIC
RESEARCH IN ENGINEERING AND MANAGEMENT, vol. 07, no. 02, Feb. 2023, doi: 10.55041/ijsrem17645.
• [7] D. Pattnaik, S. Ray, and R. Raman, “Applications of artificial intelligence and machine
learning in the financial services industry: A bibliometric review,” Heliyon, vol. 10, no. 1, Jan. 2024, doi:
Thank You!