Artificial Intelligence (417)

Artificial Intelligence (417). Unit 2 – AI Project Cycle.

Content. An Introduction to Project Problem Scoping Data Acquisition Data Exploration Modeling & Evaluation.

Problem and Goals. 3. [image] See the source image.

Activity :. Planning the school farewell party Budget arrangement Date, time and venue List of Activities in the party List of event managers Responsibilities Execution of School Farewell Party Goals – Invite all guests through an invitation card Tasks – prepare or arrange for invitation cards, get the cards ready, distribute or send cards to all guests..

An Introduction to Project. A project is a piece of planned work or an activity that is finished over a period of time and intended to achieve a particular purpose..

AI Project Cycle Stages. Problem Scoping: Identifying the problem, defining project goals. Data Collection: Collecting and compiling the relevant data. Data Exploration: Arranging the information in a proper format or layout. Modeling: Applying algorithms to data to create models. Evaluation: Assessing the model's performance and analyse the output (predictions). Example: A company wants to predict customer churn(% of Customer). The project cycle involves understanding why customers leave (problem scoping), collecting customer data (data collection), analyzing spending patterns (data exploration), training a prediction model (modeling), and checking its accuracy (evaluation)..

7. 4W’s Problem Canvas. “We are now going to go through the 4Ws Problem Canvas. This canvas helps us in identifying 4 crucial parameters we need to know for solving a problem. So what are the 4Ws? It refers to Who, What, When and Why.” Who? What? Where? Why?.

8. 4Ws problem canvas. Stakeholders : As a solution or developer, we should familiar with the people who are facing the problem and the people who will be affected by the problem directly or indirectly. 1. Who? a. Who are the stakeholders? b. What do we know about them? 2. What? a. What is the problem? b. How do you that it is a problem? (is there an evidence?) 3. Where? a. What is the context/situation the stakeholders experience this problem? b. Where is the problem located? 4. Why? a. What would hold value for the stakeholders? b. How will the solution improve their situation?.

9. [image] has [have a pro blem that when / while An ideal solution would [stakeholder(s)] [issue, problem, need] [context, situation] [benefit of solution for them] W'here Why.

Problem Scoping Factors?. Time: Deadlines and project duration Resources: Manpower, training, and tools required Finances: Budgets and financial support Challenges: Identifying and overcoming potential obstacles.

Parts of Problem Scoping. Context: Understanding the environment and circumstances Need: Identifying the gap or requirement to be addressed Vision: Defining the desired outcome or solution Outcome: Specifying measurable results and impacts.

Data Acquisition. Chapter 2. 12.

Data and Data Features. Data is the biggest asset for a business, society and economy today. Three types of data features Characters - Letters and symbols Strings Numbers Three types of data types Strings – Single Character and multi Character( String) Number – integer(2344,-1243124) and floating(Real)(234.23435) Date Complex data types Mp3,mp4, jpg(JPEG), etc..

AI System and Data. Previous batting score Training Data Utilities training data AI System Predicted batting score Testing Data.

Data Quality. Relevance – Data should not be out of context. Age – Data should not be too historic or too recent. Accuracy – Data should be correct and in proper format. Volume – Higher volume of data is better for training of machine. Richness – Variety of data values in the data set. Format – different data format is better for training of machine. Data source – depends on sources..

Data Acquisition. Definition: The process of gathering data for the AI project Two key aspects: Data Sources Data Acquisition Process.

Data Sources. Databases: Structured information storage Documents: Physical or digital files Web content: Online information Live data: Real-time data streams Raw and flat files: Unprocessed data Software applications: Data from various programs.

Data Acquisition Process. Database software: Extracting data from organized systems Document scanning: Converting physical documents to digital format Web scraping: Automatically collecting data from websites API integration: Accessing data through application programming interfaces Sensor data collection: Gathering information from IoT devices Manual data entry: Input of data by human operators.

System Map of an AI System. 19. A system map is a tool to show the relationship among various elements of a problem area in a graphical form. System map shows the interconnections of elements of a system and help us understand the complex issues easily. How much time – shorter time delay or longer time delay. How elements are related to each other – directly or indirectly. + sign : if one element increases the other will increase too. - sign : if one element increase the other will decrease..

20. [image] Clouds Oceans Underground Soil Rivers Trees Land ooo ooo Animals.

21. [image] La Su ce ive cea Clouds.

Data Exploration. Data exploration is used to obtain basic understanding of data to determine it’s suitability for training AI algorithm..

Data Exploration. Structured and Unstructured data Structured or symmetrical documents (organized manner like spreadsheet) Unstructured or does not have predefined (not organized manner like photographs, map, video clip, audio clip) Data exploration and missing values incomplete information given by person, missing values in the database Data exploration and information Data visualization – graphical representation Visualizing data for various requirements Comparing the values Establishing relationships Analyzing Distribution and composition Data Visualization Tools.

Modeling. As we enter the world of modelling, it is a good time to clarify something many of you may be having doubts about. You may have heard the terms AI, ML and DL when research content online and during this course. They are of course related, but how? Artificial Intelligence, or AI for short, refers to any technique that enables computers to mimic human intelligence. An artificially intelligent machine works on algorithms and data fed to it and gives the desired output..

25. Modelling…. Machine Learning, or ML for short, enables machines to improve at tasks with experience. The machine here learns from the new data fed to it while testing and uses it for the next iteration. It also takes into account the times when it went wrong and considers the exceptions too. Deep Learning, or DL for short, enables software to train itself to perform tasks with vast amounts of data. Since the system has got huge set of data, it is able to train itself with the help of multiple machine learning algorithms working altogether to perform a specific task. Artificial Intelligence is the umbrella term which holds both Deep Learning as well as Machine Learning. Deep Learning, on the other hand, is the very specific learning approach which is a subset of Machine Learning as it comprises of multiple Machine Learning algorithms..

26. [image] Artificial Intelligence Machine Learning Deep Learning.

27. [image] Generally, A1 models can be classified as follows: Machine Learning Learning Based Deep A1 Models Learning Rule Based.

28. [image] eese Labelled Dataset Machine identifies the image as APPLE Machine Trained using Labelled Dataset OUTPUT Rule-based A1 Model Testing Data.

Two types of AI modelling Approaches. Rule-Based Approach: Predefined rules and logic Suitable for well-understood domains Limited adaptability Learning-Based Approach: Data-driven model creation Adapts to new patterns and trends Requires large amounts of quality data.

Decision Tree. The basic structure of a Decision Tree starts from the root which the point where the decision tree starts. From there, the tree diverges into multiple directions with the help of arrows called branches. These branches depict the condition because of which the tree diverges. In the end, the final decision is where the tree ends. These decisions are termed as the leaves of the tree. You would realize that it looks like an upside-down tree..

[image] Jueunelsal o se,k sanean •eSJng e ON deelS 09 ON sax éÅÆunq I wv.

Model Evaluation. Purpose: Assess the performance and effectiveness of the AI model Key metrics: Accuracy: Correct predictions / Total predictions Precision: True positives / (True positives + False positives) Recall: True positives / (True positives + False negatives) F1 Score: Harmonic mean of precision and recall.

Ethical Considerations in AI Projects. Data privacy and security Bias and fairness in AI models Transparency and explainability of AI decisions Societal impact of AI applications Responsible AI development and deployment.