Exploring Methods and Systems of AI

Content:

• Genetic Algorithm
• Expert System
• Hopfield Neural Network

Genetic Algorithm 🧬

Task: ✈️

Given a list of cities, what is the shortest possible route that visits each city exactly once and returns to the origin city?

Functionality:

• Collecting cities to travel on the map of Ukraine 💙💛
• Storing and loading dataset of points from the MongoDB
• Tweaking genetic algorithm with different parameters 🤓
• Client can track current state of the server's work and dynamically draw a current shortest path 🖼️
• All statistic stored in MongoDB
• Server gather statistic and return a line chart on a request 📈

Built With

Backend part:

• Java 21
• Spring Boot
• Vector API. Preview feature. SIMD machine.⚡
• VirtualThreadPerTaskExecutor 🤖
• StringTemplate ➕
• MongoDB
• Integration/Unit testing
• Lombok 🏝️
• JFreeChart 📈

Frontend part:

• React 🚀
• TypeScript 💪
• HTML 5️⃣
• SCSS 😎

Preview:

Let's solve 32 points using a genetic algorithm. It would be 31! steps 🤯 without such a method.

Expert System 👩‍🔬

Task: 🧪

Create a versatile rule-based expert system capable of providing automated expert-level decision support across diverse domains. Easily extendable with new rules.

Functionality:

🎓Rule Management: Easily add new rules to the knowledge base and establish connections between existing and new rules.

🧠Rule Types: Supports various rule types, including yes/no and formula-based rules. Formula-based rules allow experts to define formulas with variables, and users can input values for these variables, enabling dynamic decision-making.

🔍Keyword Search: A powerful search feature enables users to find specific rules by entering keywords, offering a search experience similar to Google.

💬Interactive Chat: Engage in chat conversations with the system, providing user input for decision-making. The system dynamically processes the information provided by the user and offers expert-level decisions based on the rules in its knowledge base.

Built With

Backend part:

• Java 21
• Spring Boot
• Neo4j
• GraalVM ScriptEngine. It is used to calculate the formulas provided by the expert.
• Lombok 🏝️
• Integration/Unit testing 🤯

Frontend part:

• React 🚀
• TypeScript 💪
• HTML 5️⃣
• SCSS 😎

Preview:

Expert mode

User mode

Hopfield Neural Network 🧠

Unveil the potential of the Hopfield Neural Network for pattern recall, whether from images or direct matrix representations. This versatile neural network, inspired by human memory functions, excels at recognizing patterns embedded within various data forms.

Functionality:

🖌️ Drawing within a Front-end Page: Users can draw patterns on the webpage or submit matrix representations directly to the server for processing.

📷 Image and Matrix Processing: Receive images or matrix data from clients, convert images to grayscale, and process matrices to recognize embedded patterns.

👀 Pattern Recognition: Submit fragmented or incomplete patterns, either in image or matrix form, for restoration. Simply specify the network name to trigger pattern retrieval from PostgreSQL for efficient reconstruction.

Built With

Backend part:

• Java 21
• Spring Boot
• PostgreSQL 🐘
• Lombok 🏝️

Frontend part:

• React 🚀
• TypeScript 💪
• HTML 5️⃣
• SCSS 😎

Preview:

Image processing:

Create network1:

Fragment of pattern1:

Recalled pattern1:

Fragment of pattern2:

Recalled pattern2:

Create network2:

Fragments of letters:

Recalled letters:

Drawn patterns:

You can easily draw patterns on the webpage.

Create network:

Fragments of letters:

Recalled letters: