Flashcards
SCC 361
Cards in this deck (45)
- 1What is Machine Learning?A machine learns whenever it changes its structure, program, or data in such a manner that its expected future performance improves
- 2Who wrote and when was the first program that learnt to play checkers?Arthur Samuel in 1959
- 3What did Tom Mitchel say in 1998?A program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E.
- 4What are the 5 Applications of Machine Learning and give an example?Speech Recognition - Alexa, Siri, Google
Face Recognition - Facebook tag a friend suggestion
Fraud Detection - Paypal
Travel Shortest Path - Google Maps, Uber
Recommendations - Netflix, Spotify - 5What are the three things a Learning process begins with?Observation/data
Finding Patterns
Make Better Decisions - 6What is training data?This is data that is used by Machine Learning Algorithms to learn from and to build a model
- 7What is Testing Data?Data the model has never seen
- 8What does cross-validation help to prevent?Cross-Validation helps to prevent overfitting
- 9What is Overfitting?This is where the model fits with a particular set of training data and this means that it fails to fit additional data or predict future observations reliably
- 10What are the three types of Machine Learning Algorithms and give an example of each?Supervised - Linear Regression, Logistic Regression, Naive Bayes, KNN, Decision Trees, Neural Networks
Unsupervised - Clustering
Semi-supervised - A combination of the both - 11What is a Feature and give 3 examples?Individual measurable property or characteristic of a phenomenon being observed.
Examples include: Shape, Colour, Texture, Taste, Price, Origin - 12What does having irrelevant features in a model lead to?This can lead to the decrease in accuracy of a model
- 13What are the three benefits of Feature Reduction and explain each?Reduces Overfitting - Less redundant data means less opportunity to make decisions based on noise
Improves Accuracy - Less misleading data means modelling accuracy improves
Reduces Training Time - Less data means algorithms train faster - 14What is Text Classification and give an exampleThis assigns one or more classes to a document according to its content
Examples include: Detecting spam emails, Detecting news articles topics, Assist search engines - 15What is the Original "Imitation Game"Interrogator, assisting agent, deceptive agent
Determine which agent is male and which is female
A machine is intelligent if it can fool the interrogator as often as a human - 16What is the Modern Turing Test?You have only two agents involved
One agent has to determine whether the agent is a human or a machine - 17What are the three responses to the Chinese Room?Systems Reply - Searle doesn't understand, but the room as a whole does
Robot Reply - What is missing is a grounding of the symbols to real-world objects
Brain Simulator Reply - Rule-matching program actually represents neural firing - 18What is the Chinese Room?This is where you have someone in a room reading a book of Chinese letters and is able to translate them (they look like they know Chinese but they don't)
- 19What does CAPTCHA stand for?Completely Automated Public Turing test to tell Computers and Humans Apart
- 20What are the three Formulations in the 9 Queens Problems and what is the state space?Formulation 1 - Place a queen on any open source, repeat until all queens are placed.
State Space: 64*63*62*61*60*59*58*57= 1.78*10^14
Formulation 2 - Place a queen on any square in row 1, place a queen on any square in row 2.
State Space:
8*8*8*8*8*8*8*8 = 1.68*10^7
Formulation 3 - place a queen on any square in row 1, place a queen on a square in row 2 that is not in the same column
State Space:
8*7*6*5*4*3*2*1 = 40,320 - 21What are the 4 things that help to measure performance and explain each?Completeness - The strategy guaranteed to find a solution when one exists
Time Complexity - How long does it take to find a solution
Space Complexity - How much memory does it require to perfome the seach?
Optimality - Does the strategy find the best-quality solution when more than one solution exists - 22What is a Heuristic FunctionsIf we could identify an evaluation function, which describes how valuable each state was in obtaining the goal, then we could simply always choose to expand the lead node with the best value
- 23What is the function of the A* Search?Minimise the total path cost (f) = Actual path so far (g) + Estimate of future path to goal (h)
- 24What is Bayes Rule?P(A|B) P(B) = P(B|A) P(A)
P(B|A) = P(A|B)P(B)/P(A) - 25The 6 points in the outline of Basic Genetic Algorithms?1) Start - Generate random population of n chromosomes
2) Fitness - Evaluate the fitness of each chromosome
3A) Selection - select two parent chromosomes based on fitness
3B) Crossover - crossover the parents to form offspring
3C) Mutation - Mutate new offspring at some position in the chromosome
3D) Accepting - Place new offspring in a new population
4) Replace - use newly generated population for a further run of the algorithm
5) Test - if the end condition is satisfied, stop, and return the best solution in current population
6) Loop - go to step 2 - 26What are the four types of encoding when it comes to chromosomes and give an example of where it is used?Binary Encoding - Each chromosome has a binary string and an example: Knapsack Problem
Permutation Encoding - Each chromosome is a string of numbers, which represent numbers in a string and an example: Traveling Salesman Problem (TSP)
Value Encoding - Each chromosome is a string of values and an example: FInding weights for a Neural Network
Tree Encoding - Each chromosome is a tree of objects, such as functions or commands in a programming language and an example: Finding a Function from given Values - 27When defining a fitness function, if the correct answer is known. What is the fitness?The fitness is some "distance metric" toward the correct answer
- 28When defining a fitness function, if the correct answer is unknown. What is the fitness?The fitness must be an estimator of the value of the solution
- 29Name the three Selection Types when choosing chromosomes (One is a bit dumbly named tbh)?Easy Way - Take the n most-fit individuals/chromosomes
Roulette Wheel Selection - Parents allocated space on the wheel in proportion to their fitness, marble is then thrown and selects the chromosomes, chromosomes with larger fitness values will be selected more times.
Rank Selection - Every chromosome receives fitness from its rank, the worst will have fitness 1, the best will have fitness N - 30What are the problems when taking the easy way for selecting chromosomesYou Lose diversity in the chromosomes and you can also be stuck in the local minima (a smaller margin, chromosomes/child nodes which are similar)
- 31What is elitism in terms of choosing chromosomes?This is when you copy the best chromosomes to the new population
- 32Why is elitism used when choosing chromosomes?When creating a new population by crossover and mutation, there is a good chance that the best chromosomes will be lost
- 33What are the four crossover types in binary encoding?Single Point Crossover - You choose one point to crossover
Two Point Crossover - You choose two points to crossover
Unifrom Crossover - You choose three points to crossover
Arithmetic Crossover - Some arithmetic operation is performed to make a new offspring - 34What are the Advantages of Genetic Algorithms?Parallelism
Less likely to get stuck in local extrema than some other optimisation methods
relatively easy to implement - 35What Disadvantages of Genetic Algorithms?Choosing an encoding and fitness function can be difficult
computational time
no guarantee on a solution
Strong dependence on parameters - 36What are the three rules when on a Max node in alpha-beta pruning?Child > Value, Value == Child
Child > Alpha, Alpha == Child
Child >= Beta, Prune Rest of Branch/s (Forget other children, RIP TOM) - 37What are the three rules when on a Min node in alpha-beta pruning?Child < Value, Value == Child
Child < Beta, Beta == Child
Child <= Alpha, Prune Rest of Branches (Forget other children, RIP BETH) - 38What are the 5 Types of ambiguity?Pragmatic - Meaning in context
Semantic - Meaning of the sentence
Anaphoric - Referential, What/Who does each pronoun refer to?
Syntactic - Structure/Grammar, How do the words fit together in the sentence?
Lexical - Word Level, Word can have more than one meaning - 39What does Metonymy mean and give an example?One object stands for another.
Example: "Citroen announced a new car" - 40What is a metaphor and give an example?This is a figure of speech.
Example: "price have risen" - 41What are the 5 steps in Rule-Based POS Tagging?1) Apply most likely tag for each word
2) Improve the result iteratively by applying a set of patching rules
3) Use Immediate Context
4) Patching Rules
5) Tagger looks at the application of each rule of this type of training corpus, and computes the number of errors remedied by its application and the number of new errors introduced - 42What is Probabilistic POS Tagging?If we have a sequence of words, each with one or more potential tags. Then we can choose the most likely sequence of tags by calculating the probability of all possible sequences of tags, and then choose the sequence with the highest probability
- 43What are four different semantic type/aspects and explain each?Lexical Semantics - Study of the meanings of linguistic lexical units
Pragmatic Semantics - Meanings of language in context
Grammatical Semantics - Study meaning of words which have relevance to the grammatical role, or syntax
Logical Semantics - Study relations between natural language and formal logical systems - 44What are the four causes for word sense ambiguity?Homophony - Different words have the same sound
Homonymy - Different words have the same spelling
Polysemy - A word may have more than one sense
A Figure of Speech - A word or MWE (Multiword Expression) used in a non-literal sense - 45What are the Word Sense Relations?Synonym vs. Antonym - Similar or opposite
Hypernym vs. Hyponym - Is-A Relation
Meronym - Part-Of Relation
Information
- Description
- Degree Computer Science 3rd Year Flashcards on SCC 361, created by Robert Carr on 09/05/2018.
- Created by
- Robert Carr
- Views
- 0
- Rating