A-Level Data Technologies Past Paper - Using Data

Summary

This document is an A-Level past paper focusing on contemporary data practices. It covers topics such as AI, neural networks, expert systems, and data analytics, including tasks and exam questions.

Full Transcript

CONTEMPORARY DATA PRACTICES USING DATA LEARNING OBJECTIVES 1. To understand what is meant by Artificial Intelligence (AI) ​ 2. To understand the significance of the Turing test ​ 3. To understand the main features of neural network modelling ​ 4. To Understand the structure o...

CONTEMPORARY DATA PRACTICES USING DATA LEARNING OBJECTIVES 1. To understand what is meant by Artificial Intelligence (AI) ​ 2. To understand the significance of the Turing test ​ 3. To understand the main features of neural network modelling ​ 4. To Understand the structure of an expert system and its components STARTER What is AI? ARTIFICAIL INTELLIGENCE Artificial Intelligence (AI) is the simulation of intelligent behaviour by a computer which enables a machine to make decisions without human intervention. Artificial Narrow Intelligence or ANI, also known as weak AI, is created to solve one given problem. It is designed to perform a single function under various constraints and limitations. ARTIFICAIL INTELLIGENCE Artificial General Intelligence is known as strong AI and allows machines to apply knowledge and skills in different contexts.  Where ANI applications can run single, automated, and repetitive tasks, the objective of AGI is to create machines that can reason and think just like a human is capable of doing. ARTIFICAIL INTELLIGENCE Machine Learning is a subset of AI. This involves the usage of algorithms that can adapt without following explicit instructions. We utilise Machine Learning to enable systems to learn and improve from experience without being explicitly programmed. We make use of training data from large data sets in order to train Machine Learning algorithms. TASKS Write down answers to the following questions using your notes from last year. Give a definition of the training methods used within Machine Learning. What are three problems that AI can solve? THE TURING TEST The Imitation Game or the Turing Test, is a concept thought up by Alan Turin in his 1950 paper - “Can Machines Think?”. It is an inquiry, judged by humans, to determine whether or not a computer can fool humans into thinking like other humans.​ EXAM QUESTION A game developer has designed a chat bot that she hopes to use with in-game characters. Recommend an appropriate test that would allow the developer to assess the suitability of a chat bot to ‘talk’ to in-game characters. Your recommendation should show how the test would operate and the criteria needed to decide if the test had been passed. 6 Marks THE TURING TEST A human judge or Evaluator would be placed in one room. Whilst in this room the Evaluator would have a text-based conversation with two participants - one which is a human and the other a robot. At the conclusion of the test, the Evaluator would be asked if they were talking to a person or computer for each participant. ​If it was a computer, and the Evaluator believed it was a person than that computer would be deemed to have passed it. THE TURING TEST - SIGNIFICANCE The significance of the Turing Test, sets a standard for assessing machine intelligence​. It Shifts the focus from imitation to understanding – Previous AI tests were concerned with mimicking human behaviour. Passing the test requires genuine understanding and contextual adaptation.​ Motivates AI research – Many AI researcher aims to make systems that pass the test.​ literature. THE TURING THE TURING TEST - SIGNIFICANCE TEST Emphasises natural language processing - Language is a key marker of human intelligence, and the test challenges AI to master the subtleties of human language and conversation. Has profound philosophical implications – sparks debate about consciousness, the nature of intelligence, and the potential for machines to exhibit human-like thought processes.​ Raises public awareness about AI and its ethical considerations – e.g. through popular culture and literature. NEURAL NETWORKS Neural network modelling, often referred to as neural networks are a series of algorithms that mimic the operations of an animal brain to recognise relationships between vast amounts of data. This type of Machine Learning makes usage of interconnected nodes or neurons in a layered structure that resemble a brain. Similarly to a Brain, Neural networks are adaptive and learn from their mistakes. NEURAL NETWORKS Artificial neural networks attempt to solve complicated problems such as:​ Computer vision such as face recognition, image labelling or content moderation.​ Speech recognition e.g. convert conversations into documents or subtitle videos​ Natural language processing e.g. chatbots​ Recommendation engines e.g. converting social media posts into sales.​ NEURAL NETWORKS - KEY TERMS Neurons or Nodes Neural Networks consist of interconnected nodes, similar to the neurons in a human brain. Each individual node processes and transmits information. Layers Neural Networks and typically organised into Layers - Input Layer, Hidden Layer and Output Layer. Information flows from the Input into the Hidden in order to produce the Output. NEURAL NETWORKS - KEY TERMS Learning Algorithms Neural Networks use machine learning algorithms to adjust to responses due to bias or error. This process is called training, and it enables the network to learn patterns and relationships in the data. NEURAL NETWORKS - KEY TERMS Deep Learning Neural Networks with multiple hidden layers are referred to as deep neural networks. This method of learning is capable of handling complex tasks such as image recognition. DEEP LEARNING NEURAL NETWORKS - APPLICATION Some of the applications of Neural Networks are: Image and speech recognition​ Neural networks excel at pattern recognition.​ They can identify intricate patterns in data, making them valuable for tasks like image and speech recognition​ Natural language processing ​ (e.g. chatbots and virtual assistants)​ Autonomous vehicles​ Healthcare (e.g. disease diagnosis)​ Finance (e.g. stock market prediction) EXAM QUESTION Describe the concept of neural networks in artificial intelligence. 6 Marks WHAT MARK WOULD YOU AWARD THIS RESPONSE? Student A Neural networks are kind of like computer brains. They're made up of parts called neurons and layers. These parts work together to understand things, like recognising pictures or speech. There are many types of neural networks including Quantum neural networks, biological neural networks and genetic neural networks. WHAT MARK WOULD YOU Student B Neural networks are a part of artificial AWARD THIS RESPONSE? intelligence. They are composed of interconnected components known as neurons and organised into layers. These layers include input, hidden, and output layers, each having its specific role. Neurons use activation functions to process information and provide an output.​ Neural networks are trained using data and algorithms, which adjusts the connections between neurons to improve their performance. This training allows neural networks to recognise patterns and relationships in data, which is particularly useful for tasks like image recognition and language understanding. WHAT MARK WOULD YOU AWARD THIS RESPONSE? Student C Neural networks are primarily used for controlling household appliances and devices. They can be integrated into smart home systems to help automate tasks like turning on lights, adjusting thermostats, and operating kitchen appliances. Neural networks analyse user behaviour patterns and adapt devices' settings accordingly to enhance the convenience of daily life. They are especially helpful for optimising energy consumption in homes, making them more energy-efficient and cost-effective FEED-FORWARD A feed-forward neural network conveys information in one direction through input nodes which continues to be processed in this single direction until it reaches the output mode. These types of neural networks can have hidden layers for functionality. Often used in Facial Recognition. RECURRENT Recurrent Neural Networks are networks of nodes, where each node stores historical processes and these processes are reused in future processing. This type take of Neural Network takes the output of a processing node and transmit the information back into the network. ​ Often used in Text-to-Speech Applications. CONVOLUTIONAL Convolutional neural networks are especially beneficial for image recognition applications.​ These networks have an input layer, an output layer, and a hidden multitude of convolutional layers in between. The layers create feature maps that record areas of an image that are broken down further until they generate valuable outputs. ​ Often used in Image Processing. DECONVOLUTIONAL Deconvolutional neural networks are a type of neural network that works in reverse of convolutional ones, to detect items that might have been recognised as important under a convolutional neural network. ​ These items would likely have been discarded during the convolutional neural network execution process. ​ Often used in Image Processing, when looking at finding details that might have been forgotten.. MODULAR Modular neural networks contain several computer networks that work independently from one another. ​ This allows complex, elaborate computing processes to be done more efficiently. ​The goal is to have each module responsible for a particular part of an overall bigger problem. MODULAR In a Self Driving car, this would include: Perception Model - Used to look at the Car’s surroundings Navigation Model - Used to look at the Route Control Model - Used to take charge of acceleration, breaking and steering Safety Model - Used to monitor the behaviour of the car on the road EXPERT SYSTEMS An expert system is a type of artificial intelligence (AI) system designed to emulate the decision-making capabilities of a human expert. EXPERT SYSTEMS - KEY TERMS Knowledge Base Stores the specialised knowledge, rules and facts required to solve problems or make decisions within the chosen area. It's like the "brain" of the expert system. Knowledge acquisition system Responsible for acquiring knowledge from human experts and external sources and then translating it into a format suitable for the knowledge base. Facilitates the transfer of expertise from humans to the computer system. EXPERT SYSTEMS - KEY TERMS Expert interface and User interface Interface part of the system either for the human experts or the end user. Knowledge engineer A human expert who plays a crucial role in building and maintaining the expert system. They work with experts in the field (also called domain experts, e.g. Doctors for a healthcare/diagnosis system) to acquire the expert knowledge, create rules, and ensure the system's accuracy and effectiveness. EXPERT SYSTEMS - KEY TERMS Inference engine The reasoning component of the expert system. It uses the knowledge from the knowledge base and applies logical and probabilistic reasoning techniques to draw conclusions, make recommendations, or solve problems based on the input provided by the user. Explanatory system Provides explanations and justification for the decisions or recommendations made by the expert system. It helps users understand why a particular solution or advice was provided, increasing transparency and trust in the system. EXPERT SYSTEMS - KEY TERMS Working memory Temporary storage area used by the expert system during problem-solving (not the same as RAM). It holds relevant information, facts, and intermediate results as the system processes user queries and generates responses. Think of it as the system's notepad. Shell Refers to the software framework or environment in which the expert system is developed and operates. Examples include CLIPS, JESS and Prolog. EXPERT SYSTEMS - KEY TERMS Heuristics A heuristic system is designed to work with uncertainty and to simulate producing decisions based on experience. Heuristics captures information about accurate judgement and the ability to estimate and evaluate. Such rules are not derived from logic alone but are derived from a person's experience. EXPERT SYSTEMS - KEY TERMS Fuzzy logic Fuzzy logic is a concept that deals with uncertainty and imprecision in decision-making. In the context of expert systems, fuzzy logic allows the system to handle vague, incomplete, or uncertain information, making them more suitable for real-world applications where information is not always black and white. EXPERT SYSTEMS - USES Disease Diagnosis Expert systems can help doctors in diagnosing diseases based on patient symptoms and medical history. Supply Chain Optimization Expert systems can optimize inventory management and supply chain logistics to reduce costs and improve efficiency. Credit Scoring Expert systems can evaluate creditworthiness by considering various financial and personal factors. EXPERT SYSTEMS - USES Personalised Learning Expert systems can create personalized learning paths for students based on their strengths and weaknesses. Text Summarisation Expert systems can automatically summarize lengthy texts for easier consumption. NATURAL LANGUAGE PROCESSING Natural language processing (NLP) is like the language superpower of artificial intelligence (AI). It's all about teaching computers to understand and talk with us in the way we naturally speak and write. So, instead of forcing us to use computer language, NLP helps computers learn and talk like we do. NATURAL LANGUAGE PROCESSING A connected technology to NLP is Automated Speech Recognition (ASR). This is effectively the computers’ ears, it helps a computer system to turn natural language into text that it can understand. Phone Voice Assistants are common tech that uses these concepts. The ASR element, listens to your voice and turns your words into text. The NLP element, then takes over to understand what you meant and respond accordingly. HOW DOES THIS WORK? Step One - Speech Recognition When you talk to your phone, its built-in microphone captures your voice. The voice assistant’s ASR component listens to your speech and converts it into written text. HOW DOES THIS WORK? Step Two - Text Preprocessing The text is split into individual words or phrases. This allows for it to ascertain what each individual part of your text says. Step Three - Language Understanding The NLP element identifies what you want it to do, by analysing each individual word. This step also looks to indicate any specific details you have mentioned. HOW DOES THIS WORK? Step Four - Query Processing The system now understands the structure of your question Identifying the question word and the topic. It will also consider any previous interactions is has had with you and its relevance to your proposed question. HOW DOES THIS WORK? Step Five- Query Execution The voice assistant may make usage of internet to access any specific data required. This can mean accessing an API or Application Program Interface. This allows for it to fetch the information in order to answer your question. HOW DOES THIS WORK? Step Six - Response Generation Text-to-Speech or TTS Systems are used to generate a response in text that utilises the information gathers in the Query Execution step. TTS converts the text response into spoken words using a natural voice, this is called voice synthesis. HOW DOES THIS WORK? Step Seven - Output Making usage of Audio Output, your phone will speak the response. Thus creating an almost seamless answer to your question. AI USAGE OF BIG DATA A.I technologies, like machine learning, rely on large datasets to make decisions and predictions. AI systems collect data from various sources, including sensors, the internet, and user interactions. The data collected is complex and consists of numerous data types, such as text, images, and sensor readings. AI USAGE OF BIG DATA AI systems clean and organise data to make it suitable for analysis. This may involve removing outliers (data outside certain boundaries deemed inaccurate), handling missing values, and restructuring the data into a manageable form. AI models are trained using these large data sets. ETHICAL CONSIDERATIONS IN AI Expert systems are essentially helping people make decisions which can impact people’s lives and society. Ethically we can consider the following: Data Bias Algoirhtmic Bias Not knowning what AI can do Privacy Data Security Impact on Employment Informed Consent Legality? Accessability to Expert Systems Who makes the decisions? SOCIAL CONSIDERATIONS IN AI The use of AI and large data sets has many social implications both positive and negative. Socially, we can consider the following: Disenfranchise Minorities - AI systems that are biased in their decision making can lead to unfair outcomes for minority groups, which can further marginalize them. Discrimination - When AI systems are trained on imbalanced data, they may not perform well for groups that are underrepresented in the training data, such as women in car safety data. Removal of Ethics and Emotions - AI decisions are based on algorithms and data, which are devoid of ethical considerations or emotions, which can be both an advantage and a potential drawback. Removal of Privacy - The collection and analysis of large datasets by AI can pose a threat to personal privacy, as it may involve monitoring individuals' activities or sharing sensitive information. Large-Scale Unemployment - Automation and AI can replace human jobs in certain industries, potentially leading to unemployment or a shift in the types of jobs available. EXAM QUESTION Expert systems are used in many areas of finance, industry and society. Outline the role of a knowledge engineer in relation to expert systems. [ 4 Marks ] EXAM QUESTION Describe what is meant by the term heuristics in relation to expert systems. [ 6 Marks ] DATA ANALYTICS Data analytics are used in decision-making. It is the methodical computational analysis of data in order to discover and interpret meaningful patterns. Descriptive Analysis This method of analytics involves the summarisation and explanation of what occurred during a specific timeframe. This could be used to see the amount of views a website had in the last quarter. DATA ANALYTICS Diagnostic Analysis This method of analytics focuses more on why something happened. This can often involve hypothesising. This can help answer questions like did the weather affect the sale of ice cream last weekend? Diagnostic Analysis can involve machine learning in order to complete more advanced data analysis techniques. DATA ANALYTICS Predictive Analysis This method of analytics uses historical data to make informed decisions on future events. Often encompassing a variety of statistical techniques such as data mining, predictive modelling and machine learning. We can also use this to analyse current and historical facts to build predictive models within machine learning, in order to make predictions about future or otherwise unknown events. DATA ANALYTICS Prescriptive Analysis This method of analytics as the name suggests, involves providing recommendations for actions to be taken or prescribed. This can be used for healthcare treatment plans or financial investments. DATA ANALYTICS It is important that data and analysis is presented and adapted to the differing needs of the audience it is intended for. There are many factors that influence how you would present the data including audience's level of expertise, interests, demographics and needs. When presenting the data it is important that the content, format, and style of presentations are tailored to ensure they are clear, engaging, and relevant to the specific audience. DATA FLOW DIAGRAMS A data-flow diagram is a way of graphically representing a flow of data through a process or a system. Data flow diagrams basically show how data moves through a system, identifying where data comes from within the organisation, where it goes to, where it is stored and what processes it goes through on the way. The focus of this type of diagram is the organisation and what it does with the data. DATA FLOW DIAGRAMS External Entity This should be labelled with a description. This could be a person, department or external organisation. Data Store This is used to indicate a method of data storage. Process This is used to describe a process being performed. You make sure this is a verb followed by a noun - Calculate Tax. Data Flow These indicate the direction of the flow of data. They should be labelled with either the data element or data set DATA FLOW DIAGRAMS When creating a DFD the following rules should be External Entity This should be labelled with a description. This could be a adhered to: person, department or external organisation. A process must have at least one data flow entering a process and one exiting.  An external entity cannot provide data to another entity Data Store without a process taking place.  This is used to indicate a method of data storage. Data cannot move directly from an entity to a data repository without being processed.  Data cannot move from one repository to another without being processed.  Process A data repository must be connected to a process with This is used to describe a process being performed. You a data flow.  make sure this is a verb followed by a noun - Calculate A data repository must have at least one input flow and Tax. at least one output flow (even if it’s a message).  Data Flow External entities do not process data.  These indicate the direction of the flow of data. They An external entity must be connected to a process with should be labelled with either the data element or data a data flow.  set Data flows must not cross.  Objects cannot be duplicated; e.g. only one customer external entity is allowed. DATA FLOW DIAGRAMS ORDER ORDER ID CUSTOMER ORDER FOOD KITCHEN BILL ID FOOD ID INVENTORY

Use Quizgecko on...
Browser
Browser