Preface

These lecture notes aim to gives a short introduction into the basic ideas and concepts of artificial intelligence (AI). The approach and selection of topics reflect my experience with AI for the past 15 years. Most of my knowledge about the subject was self-taught through online courses, side projects and professional activities. Since my background was in applied physics,basis for getting into machine learning for natural and easy.

I hope that this notes will be useful for self-study and as a companion for more formal AI course offered elsewhere.

I have tried throughout to minimize rigorous mathematics thus making the note as accessible and self-contained as possible. Some relevant background material is provided through appendices or ‘narrative summaries’ within the main text, together with pointers to the literature.

Overview

AI is everywhere (overhyped)!

(a) Learning AI is easy!
(b) I am AI expert!
Figure 1: Famous Memes

Some notable examples:

Figure 2: Google Lens & Translate
Figure 3: Google Auto Captioning
Figure 4: Apple Face-ID
Figure 5: Tesla’s AI
  • virtual assistant (Siri, Google Assistant)
(a) Google Assistant
(b) Apple Siri
Figure 6: Intelligence Chatbot Asssistant
Figure 7: NVIDIA DLSS
  • Bloomberg (NLP)
Figure 8: Bloomberg NLP

Some notable examples (science):

Figure 9: First Image of the Milky Way’s Black Hole
Figure 10: Drug Discovery

my encounter:

  • geospatial intelligence: land use analysis
Figure 11: Land Clearing Detection (Urban Planning)
  • oil & gas: corrosion detection in confined space
Figure 12: corrosion detection in pipeline

  • automation: vehicle QC inspection (detect scratch, dent on body surface)

  • automation: vehicle QC inspection (detect missing screw in motorcycle assembly line)

  • safety: Personal protective equipment (PPE) detection for safety inspection

Figure 13: ppe detection
  • finding oil: salt detection
Figure 14: salt detection
Figure 15: General goal of AI1

Artificial intelligence, Machine learning and Deep learning

Data science and statistics - are two of the same, except that in earlier days, Data Science as we know it today, was called “statistical data analysis” or “applied statistics”.

“Data Scientist” means a professional who uses scientific methods to liberate and create meaning from raw data.

“Statistics” means the practice or science of collecting and analyzing numerical data in large quantities.

There are no real difference between the two, except that “Data Scientists” prowes in large scale data or Big Data and fast computing. Otherwise, they are the same.

Today, there are no difference between the two.2

Figure 16: Everything everywhere all at once3
Figure 17: Artificial intelligence, machine learning, and data science.4

The central goal of AI is to provide a set of algorithms and techniques that can be used to solve problems that humans perform intuitively and near automatically. A great example of such a class of AI problems is interpreting and understanding the contents of an image – this task is something that a human can do with little-to-no effort, but it has proven to be extremely difficult for machines to accomplish.

How do we relate it all?

Deep learning is a subfield of machine learning, which is, in turn, a subfield of artificial intelligence (AI).

Machine learning tends to be specifically interested in pattern recognition and learning from data. Artificial Neural Networks (ANNs) are a class of machine learning algorithms that learn from data and specialize in pattern recognition, inspired by the structure and function of the brain.

Deep learning is an approach to AI. It is a type of machine learning, a technique that allows computer systems to improve with experience and data.

Data Scientist vs Machine Learning Engineer

Figure 18: Domain area of deep learning5

MLOps is the process of automating and productize machine learning applications and workflows.

Reality: ML in production

Machine learning in production is very complicated!

Figure 19: Only a small fraction of real-world ML systems is composed of the ML code, as shown by the small black box in the middle. The required surrounding infrastructure is vast and complex.6

In a perfect world, data scientist will do ML modelling while ML Engineer will productize ML model from Data Scientist. In reality (especially in Small & Medium Enterprise), Data Scientist & ML Engineer job scope is intertwine (or even the same person!)

Figure 20: Categories of job titles related to AI
  • Formulate a definition of intelligence. How do we test this intelligence empirically ( testable and verifiable by observation)?

References

1.
Google. Google i/o 2023. (2023).
2.
Donoho, D. 50 years of data science. Journal of Computational and Graphical Statistics 26, 745–766 (2017).
3.
Urbanowicz, R. Proposed field/term venn diagram. (2018).
4.
Kotu, V. & Deshpande, B. Chapter 1 - introduction. in Data science (second edition) (eds. Kotu, V. & Deshpande, B.) (Morgan Kaufmann, 2019).
5.
Prabhu. The rise of MLOps: Managing machine learning at scale. (2023).
6.
Sculley, D. et al. Hidden technical debt in machine learning systems. in Advances in neural information processing systems (eds. Cortes, C., Lawrence, N., Lee, D., Sugiyama, M. & Garnett, R.) vol. 28 (Curran Associates, Inc., 2015).