AI Technologies

AI Technologies
• Drawing on big data analytics the use of AI technologies can benefit society on big data analytics the use of AI
technologies can benefit public policy in many ways:

 provide more accurate information, forecasts and predictions leading to better outcomes
 produce a positive social impact by using AI to provide solutions for some of the most challenging social
problems
 simulate complex systems that allow to experiment with different policy options and spot unintended
consequences before committing to a measure
 provide personalised public services
 automate repetitive and time-consuming tasks which free up
• valuable time for frontline staff

Human & Machine: AI in Public Service

https://www.youtube.com/watch?v=ppwwJ4cQQNE
- https://www.gov.uk/government/collections/a-guide-to-using-artificial-intelligence-in-the-public-sector
- https://research.aimultiple.com/ai-government/
- Pi, Y. (2021). Machine learning in Governments : Benefits , Chal- lenges and Future Directions. JeDEM - EJournal of EDemocracy and Open Government, 13(1), 203–219.
https://doi.org/https://doi.org/10.29379/jedem.v13i1.625
Five Core AI Technologies

AI Technologies

Natural Language
Machine Learning Computer Vision Robotics Deep Learning
Processing
 AI Technology – Machine Learning
• Machine Learning (ML) algorithms …
• are the “methods that can automatically detect patterns in data, and then … use the uncovered patterns to predict
future data or other outcomes of interest” (Murphy, 2012)
• is the most widely-used form of AI, and has contributed to innovations like self driving cars, speech
recognition and machine translation etc.

• Recent advances in machine learning are the result of:

• improvements to algorithms
• increases in funding
• advances in ML coincide with the ever-growing amount of “big data” collected by government agencies and
private companies or voluntarily submitted by individuals
• increased access to computational power and the expansion of cloud computing

- Murphy, K. P., Machine Learning: A Probabilistic Perspective, MIT Press, 2012. https://www.amazon.com/Machine-Learning-Probabilistic-Perspective-Computation/dp/0262018020
- https://www.gov.uk/government/collections/a-guide-to-using-artificial-intelligence-in-the-public-sector
 Machine learning algorithms

• Rashidi, H. H., Tran, N. K., Abb, H., Betts, E. V., Howell, L. P., & Green, R. (2019). Artificial Intelligence and Machine Learning in Pathology : The
Present Landscape of Supervised Methods. 6. https://doi.org/10.1177/2374289519873088
 Machine learning algorithms: Supervised learning
• Supervised learning - employ “labeled” datasets for model training and then classify unlabeled input data or
predict outcomes based on its prior training parameters (testing).

• By "labeled" data we mean:

• data where an end outcome or answer are known for previous decisions or
• the result of manual processing by experts of the raw data adding one or more meaningful and informative labels
to provide context.
• For example, we may label what is shown in the photo

• Supervised learning algorithms:

• Classification
• Neural Network Models for Human Activity Recognition for Senior Citizens

• Regression
• Decision Trees model for predicting and preventing customer churn

- Rashidi, H. H., Tran, N. K., Abb, H., Betts, E. V., Howell, L. P., & Green, R. (2019). Artificial Intelligence and Machine Learning in Pathology : The Present Landscape of Supervised Methods. 6.
https://doi.org/10.1177/2374289519873088
- Models for machine learning https://developer.ibm.com/articles/cc-models-machine-learning/
- https://www.projectpro.io/article/neural-network-projects/440
- Baduge, S. K., Thilakarathna, S., Perera, J. S., Arashpour, M., Sharafi, P., Teodosio, B., Shringi, A., & Mendis, P., Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods
and applications, Automation in Construction, Volume 141, 2022, 104440, https://doi.org/10.1016/j.autcon.2022.104440
 Machine learning algorithms: Unsupervised
learning
• Unsupervised learning - identifying hidden structures within unlabeled datasets based on the features
similarity and resulting in groups or clusters.

• Labels of clusters (groups) are not provided; hence, the right output is not known
• It becomes difficult to determine the results accuracy
• Used mostly in an exploratory way, e.g., for better comprehension of data

• Unsupervised learning algorithms:

• Clustering
• K-means algorithm for grouping retail customers to find subgroups with specific spending habits

• Dimensionally reduction
• Principle Component Analysis for identify push and pull factors affecting migration

- Rashidi, H. H., Tran, N. K., Abb, H., Betts, E. V., Howell, L. P., & Green, R. (2019). Artificial Intelligence and Machine Learning in Pathology : The Present Landscape of Supervised Methods. 6.
https://doi.org/10.1177/2374289519873088
- Models for machine learning https://developer.ibm.com/articles/cc-models-machine-learning/
- Baduge, S. K., Thilakarathna, S., Perera, J. S., Arashpour, M., Sharafi, P., Teodosio, B., Shringi, A., & Mendis, P., Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods
and applications, Automation in Construction, Volume 141, 2022, 104440, https://doi.org/10.1016/j.autcon.2022.104440
 Machine learning algorithms: Reinforcement
Learning
• Reinforcement Learning - finds optimal solutions to sequential decision problems
through trial and error.

• An example, International Business Machine (IBM)’s Deep Blue (Armonk, New

York) and Google’s Go (Alphabet, Mountain View, California) were able to beat
Chess and Go champion players respectively
• Are often flexible and adaptable enough to determine context and resolve
ambiguities
• Are often better suited to help solve problems in real-world situations

• Reinforcement Learning algorithm:

• Policy (Markov)
• Deep reinforcement learning approaches for global public health strategies for
COVID-19 pandemic

- Murphy M. Google’s AI Just Cracked the Game that Supposedly No Computer Could Beat. Quartz; 2016. https://qz.com/603313/googles-ai-just-cracked-the-game-that-supposedly-no-computercould-beat
- Greenmeier L. 20 Years After Deep Blue: How AI has Advanced Since Conquering Chess. Scientific American; 2017. https://www. scientificamerican.com/article/20-years-after-deep-blue-how-ai- has-advanced-since-conquering-chess
- Rashidi, H. H., Tran, N. K., Abb, H., Betts, E. V., Howell, L. P., & Green, R. (2019). Artificial Intelligence and Machine Learning in Pathology : The Present Landscape of Supervised Methods. 6. https://doi.org/10.1177/2374289519873088
- Baduge, S. K., Thilakarathna, S., Perera, J. S., Arashpour, M., Sharafi, P., Teodosio, B., Shringi, A., & Mendis, P., Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods and applications,
Automation in Construction, Volume 141, 2022, 104440, https://doi.org/10.1016/j.autcon.2022.104440
 AI Technology – Machine Learning Applications

Machine Learning ML Overview Algorithms Use Cases and Applications. https://www.slideteam.net/machine-learning-ml-overview-algorithms-use-cases-and-applications.html

Natural Language Processing (NLP)
• NLP is …
• syntax-based analysis of unstructured data sources
• a subfield of artificial intelligence and computational linguistics
• the studies of "understanding" the natural human language

• What is “Understanding”?
• Human understands, what about computers?
• Natural language is vague, context driven
• True understanding requires extensive knowledge of a topic
• Can/will computers ever understand natural language the same/accurate way we do?
 Natural Language Processing Task Categories (1)
• Automatic summarization - condensing a piece of text to a shorter version preserving key
informational elements and the meaning of content
• extractive - picks up sentences directly from the document based on a scoring function to form a
coherent summary
• abstractive - aim to create a summary by interpreting text and sentence paraphrasing using advanced NLP
techniques to create new shorter text
• Interactive Question answering - retrieving valid and accurate answers to user question asked in
natural language instead of query
• the answer can be extracted from a portion of a document or generated by condensing/summarizing
fragments
• Automatic speech recognition - converts spoken words into computer texts (Siri,Apple, Alexa, etc)
• could be used by chatbots to provide alternative ways to interact with end-users

- https://www.analyticsvidhya.com/blog/2021/01/introduction-to-automatic-speech-recognition-and-natural-language-processing/
- Natural Language Processing for Public Services DIGIT Directorate-General for Informatics. https://joinup.ec.europa.eu/collection/catalogue-services/document/study-natural-language-processing-public-services
- https://joinup.ec.europa.eu/interoperable-europe
- https://medium.com/luisfredgs/automatic-text-summarization-with-machine-learning-an-overview-68ded5717a25
 Natural Language Processing Task Categories (1I)
• Machine translation
• is applied in the creation of online translation apps, B2B domain-specific translation, speech-to-speech
translation, and other areas
• Natural language generation & understanding
• e.g., automated journalism, where a computer searches the Internet for real-time news, extracts data from
various sources, and writes a text summary
• Speech-to-text applications
• can be used instead of handwritten dictation, for example, to transcribe minutes of meetings
• Text-to-speech apps
• can be used to read books, articles, or other text aloud, or to add narration to videos
• Optical character recognition – extracts characters from images (for example, scan a form or a receipt)
and turn them into machine- and human-readable text

- ttps://medium.com/luisfredgs/automatic-text-summarization-with-machine-learning-an-overview-68ded5717a25
- https://joinup.ec.europa.eu/interoperable-europe
- https://www.digital-adoption.com/optical-character-recognition/
 Natural Language Processing Applications
Marketing applications
 Enables better CRM
Medicine and biology
 Literature-based discovery
Academic applications
 Research stream analysis
Healthcare
 Prioritising improvement issues
Justice
 Preparing judicial opinions
Hiring and Recruitment
 Identifying the fit and not-fit candidates
Digital businesses
 Target Advertising and gain potential customers

https://appinventiv.com/blog/natural-language-processing-applications-for-business/
https://www.mygreatlearning.com/blog/trending-natural-language-processing-applications/
Interactive Question Answering Application –
ChatGPT Video Clip

22 Jan. 2019
Source - Titel of the presentation Page 14
https://www.youtube.com/watch?v=qqxdsooKggE
Computer Vision
• Computer Vision …

• refers to the ability of AI to process and

synthesize visual data (e.g. detect and classify
objects based on images or videos)

 perform tasks such as facial recognition and

scene interpretation

 some applications involve the construction of

visual data, such as the creation of 3D models
from 2D images

Berryhill, J., & Clogher, R. (2019). Hello, World : Artificial intelligence and its use in the public sector (Issue 36).
Computer Vision Applications
• Medicine
 detection of diseases such as cancer
• Education
 image or video description and captioning (with NLP )
• Transport
 scanning license plates enabling cashless road tolls
• Facial recognition
 person identification, recruitment and hiring, policing (with ML)
• Trading
 trademark search enables users to upload a logo and return results based on visual similarity (with
ML)
• Public safety
 surveillance systems to detect possible threats to public safety and order in real-time (with ML)

Berryhill, J., & Clogher, R. (2019). Hello, World : Artificial intelligence and its use in the public sector (Issue 36).
 Robotics
• Robotics is ….
 a separate entity in AI that helps study the creation of intelligent robots or machines
 combines electrical engineering, mechanical engineering, and computer science & engineering as they have
mechanical parts, electrical components and are programmed with programming language

• What is a robot?
 A robot is a machine that (looks like a human and) can perform out-of-box actions and replicate certain human
movements automatically by means of commands given to it using programming

• AI technologies used in Robotics

• Computer Vision
• Natural Language Processing
• Reinforcement Learning

Human-like robot Sophia

- Robotics and Artificial Intelligence. https://www.javatpoint.com/robotics-and-artificial-intelligence
- https://www.educba.com/robotics-and-artificial-intelligence/
- https://www.hansonrobotics.com/sophia/
 Robotics Applications
• Defense sectors
• anti-submarine operations, fire support, battle damage management, strike missions
Medical sectors
• laparoscopy, neurosurgery, orthopedic surgery, disinfecting rooms, dispensing medication
Industrial Sector
• cutting, welding, assembly, disassembly, pick and place for printed circuit boards, packaging & labelling,
palletizing, product inspection & testing, colour coating, drilling, polishing and handling the materials
Entertainment
• camera-wielding robot for movies, animation, games and cartoons
Mining industry
• robotic dozing, excavation and haulage, robotic mapping & surveying, robotic drilling and explosive
handling
• navigating flooded passages and using cameras and other sensors to detect valuable minerals
• excavation to detect gases and other materials and keep humans safe from harm and injuries

Robotics and Artificial Intelligence. https://www.javatpoint.com/robotics-and-artificial-intelligence

The “Sophia Robot” Example – Video Clip

Source:
https://www.youtube.com/watch?v=S5t6K9iwcd
 Deep Learning
• Deep Learning is ….
 is inspired by the biology of human brains
 is a subset of ML that's based on Artificial Neural Networks (ANN)
 require computers with powerful processing, computational capabilities and Biologics vs Artificial Neural network

large amounts of data to use for training

 needs lower human intervention

• What is Artificial Neural Networks?

 scientists estimate that there are up to 100 billion neurons in the human brain
 neurons are connected to each other by synapses that pass on information by
sending electrical impulses back and forth, in the process “exciting” or
“activating” the neurons
 Artificial Neural Networks try to replicate these mechanisms and behaviours
using math

Berryhill, J., & Clogher, R. (2019). Hello, World : Artificial intelligence and its use in the public sector (Issue 36)
https://towardsdatascience.com/intro-to-deep-learning-c025efd92535
https://builtin.com/artificial-intelligence/ai-vs-machine-learning
https://sites.google.com/site/mrstevensonstechclassroom/hl-topics-only/4a-robotics-ai/neural-networks-computational-intelligence?overridemobile=true
 Deep Learning Application
• Autonomous vehicles
• robotaxi vehicles learn from some of the most challenging driving situation (with Computer Vision)
• Agriculture
• lettuce bot that knows how to single out weeds for chemical spraying
• CRM systems
• predictive lead scoring to identify customers with best chances to close
• Fraud detection
• detect anomalies in user transactions
• Vocal AI
• to identify and suppress background noise in audio files
• Healthcare
• to detect skin cancer through images

How AI Helps Autonomous Vehicles See Outside the Box

https://www.youtube.com/watch?v=HS1wV9NMLr8
Ameri A. A Deep Learning Approach to Skin Cancer Detection in Dermoscopy Images. J Biomed Phys Eng. 2020 Dec 1;10(6):801-806. doi: 10.31661/jbpe.v0i0.2004-1107. PMID: 33364218;
PMCID: PMC7753251
https://www.mygreatlearning.com/blog/deep-learning-applications/
https://builtin.com/artificial-intelligence/deep-learning-applications
https://glweb.eu/blog/digital-transformation/155-top-fascinating-application-of-deep-learning
Summary of Key Points
• Five key AI technologies include Machine Learning, Natural Language Processing, Computer Vision,
Robotics and Deep Learning
• There are three basic types of machine learning algorithms including supervised, unsupervised and
reinforcement learning algorithms.
• Key Natural Language Processing tasks are summarization, interactive question answering, automatic
speech recognition, machine translation, etc.
• Application areas for computer vision include public safety, transportation, facial recognition, etc,
• Robotics are used in different industries including manufacturing, mining and entertainment
• Deep learning is used in autonomous vehicles along with computer vision.