Build Your Own AI Classifier with Near Pocket

Learn the Basics of Artificial Intelligence

Artificial intelligence (AI) might sound complicated, but this guide will help you understand its core ideas. You’ll learn what AI is, how machine learning works, and why using AI responsibly is important. We’ll use a fun, hands-on tool called Near Pocket to build your very own AI image classifier. By the end, you’ll know how AI tools learn and how to think critically about them.

Prerequisites

• A device with a webcam (laptop, tablet, or phone).
• Internet access to use the Near Pocket website, or Android/Windows apps if you prefer offline access. (iPhone users will need to use the website).
• An optional project worksheet or a piece of paper to write down your thoughts.

Understanding Artificial Intelligence

AI is about creating technology that can do tasks that usually require human intelligence. Think about what makes humans smart. Two key ideas are autonomy and adaptivity.

Autonomy and Adaptivity

Autonomy means being able to act and make decisions on your own. Adaptivity means being able to learn from experiences and get better over time. AI systems also show these qualities.

• AI Autonomy: Technology that can help or work by itself.
• AI Adaptivity: Technology that improves by looking at more information.

AI vs. Not AI

Consider a search engine like Google. When it suggests words as you type, that’s autonomy. It’s also adaptive because it learns from what people search for to give better suggestions. This is AI.

A basic calculator, however, is not AI. It’s not autonomous because it only does exactly what you tell it to do. It’s also not adaptive; it can’t learn from new data and always works the same way.

How AI Helps People

AI tools can help us in many ways, like improving our work, doing parts of our tasks so we can focus elsewhere, or making things simpler. For example, farmers use AI to check if plant leaves are sick by taking a picture. Doctors use AI to help spot tiny signs of illness in X-rays. AI can even help by taking notes during patient visits so doctors can focus on talking to patients.

Expert Note: Remember that AI is a tool. People decide how to use it for the things they care about.

Designing Your AI Classifier: Project Time!

Now it’s your turn to create an AI classifier. You’ll design it over the next few lessons. You can choose any theme, like different types of fruit, school supplies, or even sign language letters.

Lesson 1: Choose Your Theme

Think about what you want your AI to recognize. For example, a student chose to classify fruits because they love them and they are easy to find.

On your worksheet or paper, write down what theme you choose and why.

Understanding Machine Learning

Machine learning is a way for computers to learn without being told exactly what to do for every single step. It’s like how humans learn.

Human Learning vs. Machine Learning

When you study for a test, you gather information (like notes), you study it, and your brain connects this new information to what you already know. This makes you better prepared. Machine learning works in a similar way, but uses math to find patterns in data and make predictions.

Key Parts of Machine Learning

There are four main parts:

1. Model: This is what finds the patterns.
2. Data: This is the information you give to the model to learn from.
3. Training: This is the process where the model analyzes the data to find patterns.
4. Trained Model: This is the result after training, ready to make predictions.

Just like your brain learns from notes through studying to become prepared, a model learns from data through training to become a trained model ready for predictions.

Lesson 2: Data Collection

Now, let’s think about the data for your project. What items will you classify? How will you gather pictures for them?

On your worksheet, write down the items you will classify and how you will get your data (e.g., taking pictures at home or in a garden).

Building Your AI Classifier with Near Pocket

Near Pocket is a tool that lets you add images, train a model, and see how AI recognizes things. It’s a great way to understand machine learning hands-on.

Step-by-Step: Creating Your Classifier

1. Open Near Pocket: Start a new project and give it a name (e.g., “School Supplies Project”).
2. Create Groups: You need at least two groups for your classifier. For example, you could create groups for “Scissors” and “Pens”.
3. Add Data (Images): Take pictures of items for each group. Aim for several photos per group (e.g., 6 photos of scissors, 6 photos of pens). You can add more later to improve your classifier.
4. Train the Model: Go to the “Training” section and click the “Train Model” button. This will take a few minutes.
5. Test Your Classifier: Take a new picture of an item. Your AI should now tell you if it’s a pen or a scissor. Try taking pictures from different angles.

Why AI Makes Mistakes

Your AI classifier might not always be correct. This happens because AI doesn’t understand things like humans do. It looks at every detail in a photo to find patterns.

• Limited Data: AI can only predict based on the data it has seen. If it hasn’t seen grass before, it won’t know what it is.
• Pattern Misinterpretation: Sometimes AI finds patterns that humans don’t expect. For example, if a scissor looks very similar in color and position to a colored pencil in the training photos, the AI might get confused.
• Language Limits: AI tools can only work with the languages they were trained on. If you speak to an AI in a language it doesn’t know, it might repeat a phrase it learned.

Expert Note: The quality of the data you give the AI directly affects how well it can make predictions. “What comes in, comes out.” If your data is limited or unclear, your AI’s predictions will also be limited.

How Machines Train: Algorithms

In machine learning, training is a process where the model finds patterns in data. This process follows a set of instructions, called an algorithm.

The Describer-Drawer Game

Imagine playing a game where one person describes a drawing and the other tries to draw it based only on verbal instructions. This requires clear, step-by-step directions. This is similar to how algorithms work.

What is an Algorithm?

An algorithm is simply a set of clear, step-by-step instructions to complete a task. We use algorithms every day for things like cooking, getting dressed, or solving problems.

Lesson 3: How Machines Train

Now you’ll explore how machines train. You’ll learn about algorithms and how they help machines process information.

Human Learning Algorithm

Humans learn by guessing, checking their answers, and adjusting their focus. We look for what seems important and try to understand patterns. This process involves curiosity and a desire to find meaning.

After completing these lessons, you should have a completed worksheet and your own AI classifier design. You can then present your project and your reasoning if your teacher sets up a presentation day.

Source: AI Foundations for Absolute Beginners (YouTube)