Chapter 11, Problem 2RQ

Explanation of Solution

Artificial Intelligence and major AI techniques:

• The techniques utilised to improve computer systems and make their intelligence equal or better than that of humans is called artificial intelligence.
• Human intelligence is very complex, and a single computer system cannot cover all its functionalities.
• AI systems try to accomplish certain tasks, to act and respond like humans...

Explanation of Solution

Expert System:

They are the first large scale applications of AI and account for 20% of AI systems used today. It is an intelligent technique for capturing tacit knowledge.

Working of expert systems:

• Expert systems are in a limited scope of human expertise where the skilled employees’ knowledge is collected through interviews.
• This is then transformed into a set of rules and then into a computer code.
• The knowledge collected is called knowledge base and the inference engine is the strategy used to search through the rules and make conclusions.
• These rules can be from a small number to a large volume, and inference engines help in the analysis of the same...

Explanation of Solution

Machine Learning and its working:

• Machine learning are software that can identify patterns from large databases without the need for extensive programming but need to be human trained initially.
• When analysing large databases, ML records patterns and relationships with data points and making statistical references.
• At present, around 75% of the AI utilised have some form of machine learning used.
• Machine learning can be of two types
  • Supervised learning
    • Supervised learning is where the system is being trained significantly by humans by feeding examples of desired input and output.
    • The machine learning process the examples and is then tested using a test database to ensure that the algorithm can achieve the desired output.
    • Then as the database grows, the system will improve its performance as well. An example of this includes autonomous vehicles...

Explanation of Solution

Neural networks:

• Neural networks are algorithms functioning somewhat like human neurons.
• They classify objects into known categories based on data inputs.
• They can solve complex problems, identifying patterns and relationships from massive data collections.
• They are pattern detection programs.

Deep learning:

Deep learning utilises multiple layers of neural networks to identify the underlying patterns in data, and in some cases can reveal patterns in the absence of human training.

Working of neural networks:

• The neural network can be controlled by researchers using a Learning rule.
• Collections of neurons are called nodes or layers.
• The algorithm or learning rule identifies the successful paths and strengthens the connections between the neurons comprising these pathways.
• During the analyses of the millions of available data, the process stops after attaining an accepted level of patterns being recognised.
• The learning rule from the patterns in data is then used to construct a layer of logic.
• Then inputs are processed based on the model’s experience...

Explanation of Solution

Genetic Algorithms and its working:

• They are algorithms that is loosely based on natural process of evolution and mutation.
• They control the generation, variation, and adaptation of possible problem solutions.
• They are suitable for application in product design, and industrial system monitoring.
• They can be used effectively to reduce cost and improve profits.
• They are ideal when the problem is complex and dynamic involving lot of constraints.

Intelligent Agents and its working:

• They are software programs utilised to carry out specific and repetitive tasks for users, business process and software applications, without the necessity for human intervention...

Explanation of Solution

Computer Vision Systems:

• These systems view real-world images, emulating human vision, and can extract information from them.
• Computer vision systems include image processing, pattern recognition and image undertaking.
• These systems are used in self-driving cars, drones, military applications, and industrial tools also.
• An effective example is Deep Face, developed by Facebook, to identify every photo of a Facebook user in the site and link it with that person’s Facebook account.

Natural Language Processing:

• Human language is ambiguous and meaning can vary depending on many variables.
• NLP are algorithms that allows computers to interpret and analyse natural human language.
• They can understand not only the formatted language but also instinctive use of it...