ASSEB Class 10 AI & Robotics study plan

Define AI, machine learning, deep learning and rule-based systems, and classify ten everyday products into these categories with a one-line justification each.

Distinguish supervised, unsupervised and reinforcement learning, and match five short scenario descriptions to the correct paradigm.

Explain the bias-variance trade-off and overfitting in your own words, then identify which of six described models is likely overfitting and why.

Build a one-page glossary of 25 core terms (feature, label, training set, inference, loss, hyperparameter and similar) with a plain-language definition for each.

Trace the full supervised learning pipeline from raw data to evaluation, and annotate each stage with what can go wrong.

Compute precision, recall, F1 and accuracy by hand for a small 2x2 confusion matrix, and explain when accuracy is misleading.

Describe k-fold cross-validation and a train/validation/test split, and state why a test set must stay untouched until the end.

Compare at least four algorithms (linear regression, decision tree, k-nearest neighbours, k-means) on task type, interpretability and data needs.

Identify the data types (nominal, ordinal, interval, ratio) in a supplied dataset description and choose an appropriate summary statistic for each.

Detect and describe at least five data-quality problems (missing values, duplicates, outliers, inconsistent units, class imbalance) in a described dataset and propose a fix for each.

Explain mean, median, mode, standard deviation and correlation, and state why correlation does not imply causation using a concrete example.

Outline the steps of a small data-science project from question definition through cleaning, exploration and reporting.

Classify at least eight sensors (ultrasonic, infrared, LDR, temperature, IMU, encoder, tactile, camera) by what they measure and give one robotics use for each.

Explain the sense-think-act loop and map each stage to concrete hardware and software components in a described mobile robot.

Compare DC motors, servo motors and stepper motors on control method, torque and typical use, and select one for a stated task with justification.

Describe how sensor noise and sampling rate affect robot behaviour, and propose one filtering or averaging strategy.

Explain the roles of a microcontroller, GPIO pins, ADC, PWM and a motor driver in a robot, and draw the signal path from sensor input to actuator output.

Distinguish open-loop from closed-loop control and describe how a feedback loop with an encoder corrects motor speed error.

Describe PID control conceptually, state what each term contributes, and explain what happens when the proportional gain is too high.

Identify safety practices for wiring, power supply and motor stall conditions in a student robotics build.

Explain how an image becomes pixel data and how convolution, pooling and a classification head combine in a CNN, using a described 32x32 input.

Describe the NLP pipeline from tokenisation through embedding to a classification or generation output, and give one failure mode at each stage.

Compare classical computer vision (edge detection, thresholding) with learned approaches, and state when each is preferable.

Match four described applications (face unlock, spam filter, chatbot, defect detection) to the appropriate technique and justify each choice.

Write a project brief with problem statement, users, success criteria and constraints for a small AI or robotics build.

Break the build into at least six tasks with dependencies and a realistic timeline, and identify the two highest-risk tasks.

Define an evaluation plan stating which metric proves the project works and what baseline it must beat.

Document one iteration: what was tried, what failed, what changed and why.

Analyse three ethical scenarios (biased training data, surveillance use, autonomous machine near people) using a stated framework of fairness, accountability, transparency and privacy.

Identify the main risks of an AI or robotic system across the lifecycle and propose one mitigation per risk.

Summarise the societal impact of automation on employment and accessibility with balanced arguments on both sides.

Consolidate weak areas by re-testing the two domains with the lowest practice scores and correcting every missed item.