![Literature Review
Related works Methodology Outcome Limitations
S. Russell et al. [1]
CSP Techniques-
Backtracking Search
Effective for basic
Sudoku solving-
Foundation for logic
solvers
Not scalable to larger
grids- Slow for complex
puzzles
A. Jain et al. [2]
Genetic Algorithm (GA)-
Crossover and
mutation-based
optimization
Solves most 9x9
puzzles efficiently-
Improved time
complexity
Performance sensitive
to initial population
H. Hoos et al. [3]
Simulated Annealing
(SA)- Local search with
probabilistic acceptance
Escapes local minima-
Solves difficult variants
May converge slowly-
Requires careful tuning
of parameters
16 may 2025](https://image.slidesharecdn.com/aipresentationupdatenew-250520074237-4f56a924/85/INTELLIGENTSUDOKUSOLVERWITH-AI-BASEDOPTIMIZATION-pdf-5-320.jpg)
![Literature Review (Cont.)
Related works Methodology Outcome Limitations
R. Sutton et al. [4]
Deep Reinforcement
Learning- Q-learning,
DQN-based solving
Learns general solving
policy- Can adapt to
different puzzle sizes
Needs high training
time- Requires reward
shaping
Y. LeCun et al. [5]
Neural Network
classification- Pattern
recognition for digit
inference
Can assist in digit
recognition from
handwritten Sudoku
grids
Not used for logic
solving directly](https://image.slidesharecdn.com/aipresentationupdatenew-250520074237-4f56a924/85/INTELLIGENTSUDOKUSOLVERWITH-AI-BASEDOPTIMIZATION-pdf-6-320.jpg)
INTELLIGENTSUDOKUSOLVERWITH AI-BASEDOPTIMIZATION.pdf
- 1. INTELLIGENT SUDOKU SOLVER WITH AI-BASED OPTIMIZATION PRESENTED TO D.T.M . Shahriar Sazzad Department of CSE PROJECT PRESENTATION SPRING 2025 Department of Computer Science and Engineering, Green University of Bangladesh 16 May 2025 PRESENTED BY 1. Shoaib Akter (221002353) 2. Shahanaj Akter (213902090) 3. Md. Al-Imran (221002470)
- 2. Presentation Outline ● Introduction ● Aims and Objectives ● Literature Review ● Methodology ● Results ● Conclusion ● Reference 16 may 2025
- 3. Introduction ❑ Sudoku is a widely popular logic-based number-placement puzzle. ❑ Traditional solving methods include backtracking and constraint propagation. ❑ To design an intelligent Sudoku solver using AI optimization techniques for improved speed, adaptability, and efficiency. ❑ Develop an AI-driven solver using optimization techniques (backtracking + genetic algorithms). 16 may 2025
- 4. Aims and Objectives ❑ To understand the structure and rules of Sudoku puzzles Analyze the constraints and conditions that define a valid Sudoku solution. ❑ To implement a basic Sudoku solver using traditional methods Use backtracking and constraint propagation to create a functioning base solver. ❑ To integrate AI-based optimization algorithms Apply techniques like Genetic Algorithms, Simulated Annealing, or A* Search to enhance solving efficiency and scalability.
- 5. Literature Review Related works Methodology Outcome Limitations S. Russell et al. [1] CSP Techniques- Backtracking Search Effective for basic Sudoku solving- Foundation for logic solvers Not scalable to larger grids- Slow for complex puzzles A. Jain et al. [2] Genetic Algorithm (GA)- Crossover and mutation-based optimization Solves most 9x9 puzzles efficiently- Improved time complexity Performance sensitive to initial population H. Hoos et al. [3] Simulated Annealing (SA)- Local search with probabilistic acceptance Escapes local minima- Solves difficult variants May converge slowly- Requires careful tuning of parameters 16 may 2025
- 6. Literature Review (Cont.) Related works Methodology Outcome Limitations R. Sutton et al. [4] Deep Reinforcement Learning- Q-learning, DQN-based solving Learns general solving policy- Can adapt to different puzzle sizes Needs high training time- Requires reward shaping Y. LeCun et al. [5] Neural Network classification- Pattern recognition for digit inference Can assist in digit recognition from handwritten Sudoku grids Not used for logic solving directly
- 7. Methodology Workflow: ➢ Input: Sudoku grid (text/image) ➢ Preprocessing: Image or text inputs. ➢ AI Solver: ➢ Backtracking: Fills cells recursively. ➢ Genetic Algorithm: Optimizes backtracking with fitness functions. ➢ Output: Solved grid.
- 8. Proposed Methodology Backtracking Algorithm: Implement a backtracking algorithm to solve the Sudoku puzzle. Advanced Techniques: Enhance the solver with advanced strategies like X-wing, swordfish, and jellyfish.
- 9. Results Figure - 1: HINT Figure - 2: AI SOLVE
- 10. Conclusion ❖ AI optimization methods improve solving efficiency compared to traditional techniques. ❖ Hybrid models benefit from both logical constraints and intelligent search. ❖ The system can adapt to various puzzle complexities and grid sizes. ❖ Future improvements: enhanced RL agents, mobile app integration, and OCR for puzzle input.
- 11. Reference 1. S. Russell and P. Norvig, Artificial Intelligence: A Modern Approach, 3rd ed. Upper Saddle River, NJ, USA: Prentice Hall, 2010. 2. A. Jain and S. Singh, “A Hybrid Genetic Algorithm for Solving Sudoku Puzzles,” IEEE Access, vol. 7, pp. 115106–115115, 2019. 3. H. Hoos and T. Stützle, Stochastic Local Search: Foundations and Applications, San Francisco, CA, USA: Morgan Kaufmann, 2004. 4. R. S. Sutton and A. G. Barto, Reinforcement Learning: An Introduction, 2nd ed., Cambridge, MA, USA: MIT Press, 2018. 5. Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 2015.
- 12. Thank You! Do You Have Any Questions?