01 / OBJECTIVE
What the project set out to do
Explore foundational AI ideas through small programs whose search frontiers, state transitions, and weighted decisions can be inspected directly.
02 / MY CONTRIBUTION
My part in the work
My AI lab collection brings together 18 standalone Python programs and seven Word lab documents. It covers uninformed and heuristic search, water-jug and eight-puzzle exercises, greedy route construction, and Boolean logic implemented with fixed-weight threshold neurons.
Academic lab exercises paired with code and output screenshots. Most examples use the Python standard library; seven gate scripts require NumPy. The README distinguishes the intended algorithms from the behaviour and limitations of the archived implementations.
03 / TECHNICAL APPROACH
How it came together
01Follow the search frontier
Compare breadth-first, depth-first, depth-limited, iterative-deepening, greedy best-first, and A-star demonstrations. Inspect discovery order, recorded paths, stopping conditions, and the documented deviations from standard algorithms.
02Represent states and choices
Use two-jug and three-jug exercises, an eight-puzzle demonstration, and greedy travelling-salesperson route construction to study legal actions, candidate scores, repeated states, and termination.
03Compute neural logic
Trace AND, OR, NAND, NOR, NOT, and XOR gates using weights, biases, and thresholds, including three-input gates and the intermediate neurons used for XOR. These examples demonstrate computation with fixed parameters rather than model training.
04 / OUTCOMES
What the work produced
- Organised search, puzzle, route, and neural-logic examples into a linked learning collection.
- Paired runnable Python scripts with lab documents and sample outputs.
- Documented the difference between traversal order, reconstructed paths, heuristic choices, and optimal solutions.
Educational examples with known implementation issues. The A-star and iterative-deepening exercises need corrections before general use; puzzle and route examples do not guarantee an optimal solution. Neural gate examples use fixed weights and biases.