Alps - 17/12/24 Flashcards
what is a Server
A computer network responsible for managing and sharing resources, include file, print, application
what is the Switched Hub
Acts as a single connection point for devices on a network.
It forwards data packets to destination using each Packets Address.
Most hubs support load balancing so data packets are forwarded via different network segments based on traffic patterns.
What is a Media Converter
A hardware device that allows 2 dissimilar media types to connect with each other. May involve connecting a fibre optic cable and a copper cable.
What is an IP Address
Devices that communicate using internet are assigned an unique IP Address.
Is used to identify the device from another on the internet.
Networks will route messages using IP Address of the destination.
Consists of 32 bits formatted as 4 Octets.
Numbers can range from 0-255 (per octet) in decimal.
What is a MAC Address
(Media access control address
Allocated to a network interface card when it is manufactured is unique to it.
Made up of 6 two-digit Hexadecimal numbers, separated by colons.
P2P v Client Server Network
P2P= All computers on the network have equal status.
Allows hardware and software to function without the need for any dedicated server devices.
Each computer is a supplier and consumer of resources.
Each node is in charge of its own security.
Client= Used in Large organisations.
one computer file saver will carry out file storage, backup, provide application software and printer management.
All computers connected to server via a switch or hub.
Ring Network Topology
Consists of a number of nodes connected without the need for a file saver.
each node is connected to 2 other adjacent nodes.
A token is passed from node to node.
Difficult to add a new node to this network.
Data travel for large distances can lead to problems with the data signal, this tends to degenerate and weaken.
Star Network Topology
Has a central file saver.
Each node is connected to the file saver by its own cable.
Host computer controls all communication.
If node fails, network can still operate as only affects that node.
Adding a new node is only a matter of connecting it to the hub if there is capacity.
OSI - Open System Interconnection Model
Provides a framework for data to be transferred between network computers.
Model works even if different data formats or different types of networks are communicating.
Developed by the ISO (international standards organisation)
OSI - benefits
Any hardware that meets the OSI standard will be able to communicate with any other hardware that also meets the standard (same with software).
Consumers are given a wider choice, hardware/software from any manufacturer will work together, independent of country.
Not dependent of the operating system.
Error handling in each layer.
different layers can operate automatically.
Basic Rules of OSI
Top 3 Layers are grouped together and called the ‘Application set’.
Application set: mainly concerned with controlling how various applications currently running are making use if the network.
Bottom 4 layers are called the ‘Transport set’.
Transport set: only concerned with passing info through the network.
Each layer can only communicate with the layer above or below it.
OSI - 1. Application layer
Top Layer.
Only concerned with presenting info in human friendly way.
relevant to all typed of devices that use inbuilt computer processors, e.g., Mobile Phones
OSI - 2. Presentation layer
Concerned with presenting info to the various electronic devices in the correct format.
Can implement encryption.
May compress/decompress info.
OSI - 3. Session layer
A session begins when an application wants to make a connection to a remote server.
Opens a temporary channel between the 2, to allow communication.
Can have more than one session running at a time.
OSI - 4. Transport layer
Divides info into convenient sized packets and sent on different routes.
Packets may arrive in a different order, they will be reassembled at the destination point into the order they were sent.
Layer can also check for errors.
OSI - 5. Network layer
Works out best rout for the packets to use.
Every computer has an unique IP Address to each packet.
Reads the address of incoming packets to a computer and if they are destined for that computer, it allows them through.
OSI - 6. Data Link layer
Each Packet is converted into a series of bits.
Converts incoming bits back into complete packets.
Bits can be corrupted of flipped and the layer will attempt to spot these reversals and fix them.
OSI - 7. Physical layer
Bits and Bytes have to be converted into physical effect so they can be transmitted.
This layer will convert the abstract ‘Data Bit’ inside the computer into a physical effect of some kind.
Tuning Test
Used to assess the ability of a machine to exhibit intelligent behaviour which could not be distinguished from that of a human.
Critics challenge test as a measure of intelligence as it doesn’t assess the correctness of responses, only how similar to human response.
Tuning test - natural network modelling and Ai
Computational models used in computer science to model a primitive brain.
Tuning test - Natural network modelling and Ai
Main Features
Consists of a number of artificial neurons called units.
Number of units could vary depending on the complexity of the neural network.
They are arranged in layers.
Input units will receive info from the outside world.
Hidden units will process input from the input unit.
Output provides a response from the network representing any info it has learned.
Most cases, the neural networks are fully connected.
Each connection is weighted and algorithms are used to calculate the weighted sum of any input to generate an output value.
Neural networks learn by itself, no need to explicitly program.
How does a neural network learn?
Pattens of info will enter the network via the input units, triggering layers of hidden units until they arrive at the output units.
Inputs into a unit are multiplied by the weightings.
Weighted inputs are added together and if sum is above given threshold value, that unit will trigger units connected - feedforward network.
Feedback is important - used to support learning - Backpropagation.
The output produced by the neural network is compared to the output it was meant to produce.
Expert Systems.
Knowledge based systems.
An application of AI - in this the knowledge of a human expert is made available through a computer package.
used in a narrow field of knowledge - knowledge domain.
Expert systems - key components
An expert is an experienced practitioner in a particular field, e.g., a medical specialist, repair technician or financial analyst.
Initial building stage is called knowledge acquisition.
Knowledge of human expert is programmed into the expert system and represented using rules and facts.
Result is the creation of a knowledge base - this holds knowledge about the domain, Can be stated as a series of IF-THEM-ELSE rules.
Expert systems - factual Knowledge
Factual info acquired the knowledge engineers from the human experts.
expert systems -
info about accurate judgement and ability to estimate and evaluate.
System is designed to work with uncertainty.
Extra rules and facts may be added to the knowledge base as time passes and based on feedback.
A system for consulting is know as the inference engine - this software interrogated the knowledge base and draws conclusions based on the rules, it poses questions to the user.
expert systems - user interface
Allows user to communicate with the system.
Requests from user interface to interface engine.
Request is processed by the interface engine and then returns a response to the user.
Expert Systems - the Shell
A piece of software containing the structure for creating an expert system.
An empty expert system without a knowledge base.
Expert Systems - Fuzzy Logic
Designed to approximate the way people think when they approach a problem.
An approach to computing based on ‘Degree of Truth’.
Benefits of using an expert system
Expert advice is always available.
Can be used as a training aid to increase expertise of staff.
Doesn’t get tired or over worked.
More likely cheaper than using the time of a highly paid professional human.
Can be used to harness the combined knowledge of lots of experts and the knowledge of one can be shared globally.
Limitations of using an expert system
Can make mistakes and not learn from them - humans can.
Risk of over reliance on tech -experts may become deskilled as not applying their own intuition to situations.
Human advisor may take into account special circumstances which an expert system may overlook.
Every scenario cannot be programmed - errors may be possible.
Natural language processing and voice recognition.
how do voice recognition systems work?
Refers to the combination of hardware and software systems which decide a spoken command.
First stage is the input and digestion of spoken words.
A device with a sound card is required, along with a mic.
Some applications use separate hardware devices, while some have all necessary hardware built in.
How computers recognise speech
1. Pattern Matching
words spoken by user are recognised, often used by business’ with automated switchboard.
User will be presented with questions with limited simplistic responses.
Computer will analyse the input from the user and try to match it with a list of potential sound patterns.
How computers recognise speech
2. Pattern and feature analysis
Spoken input is recorded and digitised using an ADC (analogue to digital converter).
Data is analysed and compared to a stored dictionary.
More complex input can be analysed and user isn’t limited to responses.
How computers recognise speech
3. Statistical analysis
Can apply the rules of grammar to help predict words to support speech recognition.
How computers recognise speech
4. Artificial Neural Networks (ANN)
Scientists looking at how they can be trained through the use of examples to recognise spoken input.
Natural Language processing
A branch of AI concerned with the ability of a computer to understand human interaction in its natural spoken or written from.
The Ability of a machine to analyse, understand and generate human speech
Automated speech recognition
Can form a part of NLP process.
Some applications will use speech recognition to convert input sound to text before NLP takes place.
How natural language processing is used today:
1. Spam filters
Many organisations use NLP as their first line of defence against spam.
Use NLP to extract meaning from strings of text to help identify unwanted email.
How natural language processing is used today:
2. Answering questions
Rely on users being very specific with the key words.
Companies are focusing on the use of NLP to help with the processing of natural language questions.
How natural language processing is used today:
3. Extracting information
Many organisations are using algorithmic trading as a means of managing investments.
Financial investments are controlled by technology which will evaluate news articles and extract relevant info.
How natural language processing is used today:
4. Summarising information
Information overload posses a problem.
Social media provides use of NLP to analyse info on used collected via SM to help determine their preferences.
Recently companies programmed their applications by default to access the users microphone.
How does natural language processing work.
Computers can have trouble because they try to understand the meaning of individual words, rather than the whole sentence or phrase.
Understanding can be made more difficult due to the fact may words have double meanings so applications must also have an understanding of the context.
How does natural language processing work.
1. Morphology
How words are formed and their relationship with other words.
(Morph… shape/formation of words)
How does natural language processing work.
2. Syntax
How words and sentences are put together
(Tax…together?)
How does natural language processing work.
3. Semantics
Meanings of words and group of words
(Man…one or more?)
How does natural language processing work.
4. Pragmatics
Context of spoken expressions
How does natural language processing work.
5. Phonology
The sounds associated with spoken language
Natural Language
Part-of-speech-tagging (PoS)
The first step in natural language.
Modern applications will apply a self-learning algorithm which will tag words with multiple meanings.
Applications will determine the highest occurring meaning.
Natural Language
Parse trees / diagrams
Second step.
To use knowledge derived from syntax to understand the structure of the sentence.
Algorithm will break the sentence into noun and verb phrases.
Natural Language
Semantics
Third step.
Considers the semantics of a sentence.
This stage will process the words that appear before and after the tagged word to help apply meaning to the sentance.
Natural language has room for improvement, especially in the area of pragmatics.
EVAL
natural language recognition
ADVANTAGES
No training is required in the use of application.
Increases accessibility of the application.
If combined with speech recognition, it can free users to complete other tasks.
EVAL
natural language recognition
DISADVANTAGES
Not all commands will be recognised / identified correctly all the time.
Applications are complex to create and are therefore expensive.
EVAL
voice recognition
ADVANTAGES
No training is required in the use of the application.
Accuate spelling of words for users with literacy problems.
For those with limited mobility is a more effective method of input
EVAL
voice recognition
DISADVANTAGES
Prone to interference.
User may be required to speak slower and louder.
May only recognise a limited range of voices.
