csabatini/kbai_notes.txt

## kbai_notes.txt
Ebook: https://gatech.instructure.com/courses/193244/files/folder/KBAI-Ebook

GDrive notes #1: https://drive.google.com/file/d/1T2TNnNQbfkjR1kyiwC3hgLEjTFiMgbC5/view
GDrive notes #2: https://drive.google.com/drive/folders/1f-udSsv9oMgJ8zzP_YuJo2Em0zQgrDuC?usp=sharing

---
Lesson 3: Semantic Networks
1. Knowledge representation and Reasoning using that representation is the key to problem-solving. 2. Semantic Networks are one of the many ways for knowledge representation. 3. Pathways along spreading activation networks could potentially help with memorizing and recalling solutions instead of solving them every time for new recurring problems.

---
Lesson 4: Generate and Test
This lesson cover the following topics: 1. Generate and Test is a very commonly used problem-solving method used by humans and in nature by biological evolution (similar to Genetic algorithms). 2. We need both knowledge representation and problem-solving methods together to provide reasoning to solve problems. 3. Smart generators and smart testers help prune multitude number of states that are possible due to combinatorial explosion of successor states, thereby helping solve intractable problems effeciently using limited computational resources and limited knowledge of the world as compared to dumb generators and dumb testers.
---
Lesson 5: Means End Analysis and Problem Reduction
This lesson cover the following topics: 1. The knowledge representation of Semantic networks works well with Generate and Test, Means-Ends Analysis and Problem Reduction. These are examples of Universal AI methods. 2. Means-ends analysis uses a heuristic to guide the search from the initial state to the goal state. Convergence is not guaranteed. Optimality is not guaranteed. Computational effeciency is not guaranteed. 3. Problem reduction is used along with Means-ends analysis to help overcome problems with means-ends analysis.  4. The Universal methods have weak coupling between the methods and the knowledge representation and hence are called Weak methods since they make little use of knowledge. Strong AI methods are knowledge-intensive and use knowledge of the world to come up with good solutions in an effecient manner.
Means end Analysis
Approach: For each operation
* Apply it to current state
* Calculate difference between new and goal state
* Select/prefer move that minimizes distance between new state and goal

Pros/cons
* universal AI method
* costly and no guarantee of success or efficiency
* doesn't necessarily bring us closer to goal

Problem reduction
* Given a big problem, decompose it into smaller problems that are easier to solve. And compose into a final solutions
In general, universal methods are weak
---
Lesson 6: Production Systems
Production Systems helps map percepts in the world into actions. When the production system reaches an impasse, it uses chunking to learn a new rule to overcome that impasse.
Architectures/Layers for KBAI systems
Three layers: Knowledge/Task Level, Algo Level, Hardware Level
Example: Watson
* Hardware: Physical computer
* Algo layer: Searching and decision-making for answers
* Task layer: Answering the clue based on his knowledge, searching and answering

Architecture + Content = Behavior
Functions: Perceptions -> Action
In the 2nd model, architecture doesn't change. Similar to a computer running programs, the architecture is unchanged
E.g. SOAR architecture
https://en.wikipedia.org/wiki/Soar_(cognitive_architecture)
Production rules: Captured in the procedural knowledge in SOAR's memory. Seven example rules for pitcher problem
System can reach an impass where there is no single course of action determined
"Chunking" is a learning technique to learn rules that can break an impass (between two rules), using memory
---
Lesson 7: Frames
Frames represent stereotypes of a certain concept (e.g. situation, event, etc.) and are composed of Slots and Fillers. Frames provide default values for the Slots and can inherit from one another. Frames are representationally equivalent to Semantic Nets. In other words, a frame can capture a large amount of information in an organized manner as a packet. Frames enable us to construct a theory of cognitive processing which is both bottom-up and top-down. The data in the frames generates expectations of the world in a cognitive-effecient manner.

---
Lesson 8: Learning by Recording Cases
1. Memory is as important as Learning/Reasoning so that we can fetch the answer to similar cases encountered in the past and avoid having to redo the non-trivial task of learning and reasoning, thereby saving effort. 2. A case is an encapsulation of a past experience that can be applied to a large number of similar situations in future. The similarity metric can be as simple as the Euclidean distance metric or a complex metric involving higher dimensions. 3. kNN method is one method to find the most similar case from memory for a new problem. 4. In some cases, we need to adapt the cases from our memory to fit the requirements of the new problem. In some cases, we also need to store cases based on qualitative labels along with numeric labels to make the comparison applicable for particular situations. 5. Learning by storing cases in memory has a very strong connection to Cognition since human cognition works in a similar manner by recording cases and applying them to new problems in real world by exploting the patterns of regularity in them.
* Given a new problem 'A'
* Retrieve most similar problem from memory ('B')
* Apply the solution of 'B' to 'A'
example:
* always starting with a main method for a java project
* debugging errors using previous cases
* doctor using a similar cases when determining a diagnosis
more objective: calculate the euclidean distance (x/y) and choose the nearest neighbor
Also need methods to adapt past cases to fit the new problem (this is called case based reasoning, next lesson)
---
Lesson 9: Case-based reasoning
1. In case-based reasoning, the cognitive agent addresses new problems by adapting or tweaking previously encountered solutions to new similar but not identical problems. 2. Case-based reasoning has 4 phases: 1) Case Retrieval, 2) Case Adaptation, 3) Case Evaluation and 4) Case Storage. 3. One method of Case Retrieval is kNN method. 4. 3 common ways of Case Adaptation are: 1) Model-based method, 2) Recursive case-based method, and 3) Rule-based method. 5. Designers often use heuristics for case adaptation. A heuristic is a rule of thumb that works often, but NOT always. 6. Case Evaluation can be performed through Simulation or if the cost is not high then through actual Execution. It can also be done by building a Prototype and testing it or through careful review of the design by experts. Simulation, Prototype or Execution. 7. Case Storage has 2 kinds of mechanisms to organize information for effecient retrieval: 1) Indexing/Tabular method (Linear time complexity) and 2) Discrimination Tree (Logarithmic) 8. Incremental Learning allows the addition of a new case which enables new knowledge structure to be learnt. 9. If the evaluation of a case retrieved fails, then it could be adapted and retried and if the failure continues, then we need to abandon the case. Sometimes, storing failed cases helps us anticipate future problems. Case Adaption is done using model of the world, by using rules or using recursion. 10. We do not need to store all successful cases, but yet need to store noteworthy and representative cases so that we get enough utility from the stored cases and at the same time keep the retrieval process tractable. This is also known as the Utility problem. 11. Case-based reasoning has a very strong connection with human cognition. Case-based reasoning shifts the balance of importance from Reasoning to both Learning and Memory. Case-based reasoning unifes all the 3 concepts: Learning (to acquire experiences), Memory (to store and retrieve experiences) and Reasoning (to adapt experiences to similar new problems).

Unlike recording cases, in case-based reasoning, the new problem is similar but not identical to a previous case
Two components:
* Case-based: extract something from memory and re-use it
* Reasoning: Adapt the solution from memory to fit the new problem

CBR Steps: 1) Retrieval, 2) Adaptation, 3) Evaluation (determine how well the solution fits the new problem) 4) Storage of new solution as a case
Assumptions:
* Patterns exist in the world
* Similar problems have similar solutions
Use heuristics: rules that work sometimes but not always (rule of thumb)
---
Lesson 10: Incremental Concept Learning
Incremental concept learning is intimately connected with human cognition where instead of giving a large number of examples, the agent is given one example at a time and the agent gradually and incrementaly learns concepts from those examples.

In ICL, instead of getting a large number of examples, the agent is given one example at a time and gradually learns from these examples (positive and negative)
* Generalize: Concept is expanded to include a positive example/feature
* Specialize: Concept is limited to exclude a negative example/feature
Example: Child learning about animals: concept of a cat - black cat, orange cat, dog, etc.
also, structure of a foo/blocks: AI learns about support blocks, which sides can be touching, etc
Heuristics:
require-link, forbid-link, drop-link, enlarge-set, climb-tree, close-interval
---
Lesson 11: Classification
Classification is ubiquitous and Humans continuously and constantly perform classification in day-to-day life
It can be top-down or bottom up.
Top-down: establish and refine. starting with an animal and going deeper in the hierarchy
Bottom-up controller processing/search: DJIA price rediction.
Equivalence Classes reduce the 2^n percepts into a more manageable number of concepts for the AI agent
Concept Hierarchies: e.g. Animal -> Reptile/Mammal/Marsupial, etc.
Axiomatic Concepts, Prototype Concepts, Exemplar Concepts
1. formal set of necessary and sufficient conditions (like a circle)
2. base properties that can sometimes be overridden (prototypical) - like a stool and a folding chair are both chairs
3. defined by implicit abstractions of certain examples. example: beauty could be a flower, a sunset, a painting. Very hard to define and use in AI---
Lesson 12: Logic
Logic provide the framework for formal notation/language for reasoning and inferences. In other words, logic provides a formal and precise way of reasoning. It also allows agents to reason more formally about initial and goals states and helps in planning. Deduction is term used for reasoning from causes to effects; Abduction is the term used for reasoning from effects to causes; and Induction is generating a generic rule, given the cause and its effect.

Why do we need logic?
We have knowledge base and rules. We might want the agent to prove its answer, or only present a valid solution.
Soundness: Only valid conclusions can be proven
Completeness: All valid conclusions can be proven
---
	Ebook: https://gatech.instructure.com/courses/193244/files/folder/KBAI-Ebook

	GDrive notes #1: https://drive.google.com/file/d/1T2TNnNQbfkjR1kyiwC3hgLEjTFiMgbC5/view
	GDrive notes #2: https://drive.google.com/drive/folders/1f-udSsv9oMgJ8zzP_YuJo2Em0zQgrDuC?usp=sharing

	---
	Lesson 3: Semantic Networks
	1. Knowledge representation and Reasoning using that representation is the key to problem-solving. 2. Semantic Networks are one of the many ways for knowledge representation. 3. Pathways along spreading activation networks could potentially help with memorizing and recalling solutions instead of solving them every time for new recurring problems.

	---
	Lesson 4: Generate and Test
	This lesson cover the following topics: 1. Generate and Test is a very commonly used problem-solving method used by humans and in nature by biological evolution (similar to Genetic algorithms). 2. We need both knowledge representation and problem-solving methods together to provide reasoning to solve problems. 3. Smart generators and smart testers help prune multitude number of states that are possible due to combinatorial explosion of successor states, thereby helping solve intractable problems effeciently using limited computational resources and limited knowledge of the world as compared to dumb generators and dumb testers.
	---
	Lesson 5: Means End Analysis and Problem Reduction
	This lesson cover the following topics: 1. The knowledge representation of Semantic networks works well with Generate and Test, Means-Ends Analysis and Problem Reduction. These are examples of Universal AI methods. 2. Means-ends analysis uses a heuristic to guide the search from the initial state to the goal state. Convergence is not guaranteed. Optimality is not guaranteed. Computational effeciency is not guaranteed. 3. Problem reduction is used along with Means-ends analysis to help overcome problems with means-ends analysis. 4. The Universal methods have weak coupling between the methods and the knowledge representation and hence are called Weak methods since they make little use of knowledge. Strong AI methods are knowledge-intensive and use knowledge of the world to come up with good solutions in an effecient manner.
	Means end Analysis
	Approach: For each operation
	* Apply it to current state
	* Calculate difference between new and goal state
	* Select/prefer move that minimizes distance between new state and goal

	Pros/cons
	* universal AI method
	* costly and no guarantee of success or efficiency
	* doesn't necessarily bring us closer to goal

	Problem reduction
	* Given a big problem, decompose it into smaller problems that are easier to solve. And compose into a final solutions
	In general, universal methods are weak
	---
	Lesson 6: Production Systems
	Production Systems helps map percepts in the world into actions. When the production system reaches an impasse, it uses chunking to learn a new rule to overcome that impasse.
	Architectures/Layers for KBAI systems
	Three layers: Knowledge/Task Level, Algo Level, Hardware Level
	Example: Watson
	* Hardware: Physical computer
	* Algo layer: Searching and decision-making for answers
	* Task layer: Answering the clue based on his knowledge, searching and answering

	Architecture + Content = Behavior
	Functions: Perceptions -> Action
	In the 2nd model, architecture doesn't change. Similar to a computer running programs, the architecture is unchanged
	E.g. SOAR architecture
	https://en.wikipedia.org/wiki/Soar_(cognitive_architecture)
	Production rules: Captured in the procedural knowledge in SOAR's memory. Seven example rules for pitcher problem
	System can reach an impass where there is no single course of action determined
	"Chunking" is a learning technique to learn rules that can break an impass (between two rules), using memory
	---
	Lesson 7: Frames
	Frames represent stereotypes of a certain concept (e.g. situation, event, etc.) and are composed of Slots and Fillers. Frames provide default values for the Slots and can inherit from one another. Frames are representationally equivalent to Semantic Nets. In other words, a frame can capture a large amount of information in an organized manner as a packet. Frames enable us to construct a theory of cognitive processing which is both bottom-up and top-down. The data in the frames generates expectations of the world in a cognitive-effecient manner.

	---
	Lesson 8: Learning by Recording Cases
	1. Memory is as important as Learning/Reasoning so that we can fetch the answer to similar cases encountered in the past and avoid having to redo the non-trivial task of learning and reasoning, thereby saving effort. 2. A case is an encapsulation of a past experience that can be applied to a large number of similar situations in future. The similarity metric can be as simple as the Euclidean distance metric or a complex metric involving higher dimensions. 3. kNN method is one method to find the most similar case from memory for a new problem. 4. In some cases, we need to adapt the cases from our memory to fit the requirements of the new problem. In some cases, we also need to store cases based on qualitative labels along with numeric labels to make the comparison applicable for particular situations. 5. Learning by storing cases in memory has a very strong connection to Cognition since human cognition works in a similar manner by recording cases and applying them to new problems in real world by exploting the patterns of regularity in them.
	* Given a new problem 'A'
	* Retrieve most similar problem from memory ('B')
	* Apply the solution of 'B' to 'A'
	example:
	* always starting with a main method for a java project
	* debugging errors using previous cases
	* doctor using a similar cases when determining a diagnosis
	more objective: calculate the euclidean distance (x/y) and choose the nearest neighbor
	Also need methods to adapt past cases to fit the new problem (this is called case based reasoning, next lesson)
	---
	Lesson 9: Case-based reasoning
	1. In case-based reasoning, the cognitive agent addresses new problems by adapting or tweaking previously encountered solutions to new similar but not identical problems. 2. Case-based reasoning has 4 phases: 1) Case Retrieval, 2) Case Adaptation, 3) Case Evaluation and 4) Case Storage. 3. One method of Case Retrieval is kNN method. 4. 3 common ways of Case Adaptation are: 1) Model-based method, 2) Recursive case-based method, and 3) Rule-based method. 5. Designers often use heuristics for case adaptation. A heuristic is a rule of thumb that works often, but NOT always. 6. Case Evaluation can be performed through Simulation or if the cost is not high then through actual Execution. It can also be done by building a Prototype and testing it or through careful review of the design by experts. Simulation, Prototype or Execution. 7. Case Storage has 2 kinds of mechanisms to organize information for effecient retrieval: 1) Indexing/Tabular method (Linear time complexity) and 2) Discrimination Tree (Logarithmic) 8. Incremental Learning allows the addition of a new case which enables new knowledge structure to be learnt. 9. If the evaluation of a case retrieved fails, then it could be adapted and retried and if the failure continues, then we need to abandon the case. Sometimes, storing failed cases helps us anticipate future problems. Case Adaption is done using model of the world, by using rules or using recursion. 10. We do not need to store all successful cases, but yet need to store noteworthy and representative cases so that we get enough utility from the stored cases and at the same time keep the retrieval process tractable. This is also known as the Utility problem. 11. Case-based reasoning has a very strong connection with human cognition. Case-based reasoning shifts the balance of importance from Reasoning to both Learning and Memory. Case-based reasoning unifes all the 3 concepts: Learning (to acquire experiences), Memory (to store and retrieve experiences) and Reasoning (to adapt experiences to similar new problems).

	Unlike recording cases, in case-based reasoning, the new problem is similar but not identical to a previous case
	Two components:
	* Case-based: extract something from memory and re-use it
	* Reasoning: Adapt the solution from memory to fit the new problem

	CBR Steps: 1) Retrieval, 2) Adaptation, 3) Evaluation (determine how well the solution fits the new problem) 4) Storage of new solution as a case
	Assumptions:
	* Patterns exist in the world
	* Similar problems have similar solutions
	Use heuristics: rules that work sometimes but not always (rule of thumb)
	---
	Lesson 10: Incremental Concept Learning
	Incremental concept learning is intimately connected with human cognition where instead of giving a large number of examples, the agent is given one example at a time and the agent gradually and incrementaly learns concepts from those examples.

	In ICL, instead of getting a large number of examples, the agent is given one example at a time and gradually learns from these examples (positive and negative)
	* Generalize: Concept is expanded to include a positive example/feature
	* Specialize: Concept is limited to exclude a negative example/feature
	Example: Child learning about animals: concept of a cat - black cat, orange cat, dog, etc.
	also, structure of a foo/blocks: AI learns about support blocks, which sides can be touching, etc
	Heuristics:
	require-link, forbid-link, drop-link, enlarge-set, climb-tree, close-interval
	---
	Lesson 11: Classification
	Classification is ubiquitous and Humans continuously and constantly perform classification in day-to-day life
	It can be top-down or bottom up.
	Top-down: establish and refine. starting with an animal and going deeper in the hierarchy
	Bottom-up controller processing/search: DJIA price rediction.
	Equivalence Classes reduce the 2^n percepts into a more manageable number of concepts for the AI agent
	Concept Hierarchies: e.g. Animal -> Reptile/Mammal/Marsupial, etc.
	Axiomatic Concepts, Prototype Concepts, Exemplar Concepts
	1. formal set of necessary and sufficient conditions (like a circle)
	2. base properties that can sometimes be overridden (prototypical) - like a stool and a folding chair are both chairs
	3. defined by implicit abstractions of certain examples. example: beauty could be a flower, a sunset, a painting. Very hard to define and use in AI---
	Lesson 12: Logic
	Logic provide the framework for formal notation/language for reasoning and inferences. In other words, logic provides a formal and precise way of reasoning. It also allows agents to reason more formally about initial and goals states and helps in planning. Deduction is term used for reasoning from causes to effects; Abduction is the term used for reasoning from effects to causes; and Induction is generating a generic rule, given the cause and its effect.

	Why do we need logic?
	We have knowledge base and rules. We might want the agent to prove its answer, or only present a valid solution.
	Soundness: Only valid conclusions can be proven
	Completeness: All valid conclusions can be proven
	---