What is machine learning? by EUN Partnership aisbl

Koutelida Xristina

2024-04-22 13:58:38 UTC

Imagine you're training a dog to fetch a ball. You show the dog different balls (like tennis balls, soccer balls) and tell it which one to fetch each time. After a while, the dog learns to associate each type of ball with the correct action (fetching). In supervised learning, the AI model is provided with labeled data (input-output pairs) and learns to map inputs to outputs by observing examples.

Koutelida Xristina

2024-04-22 13:59:04 UTC

Think of sorting a bag of different-colored marbles without any labels or instructions. You start grouping them based on similarities putting red ones together, blue ones together, and so on. Eventually, you notice patterns emerging, like certain shades clustering together. This is similar to unsupervised learning, where the AI model identifies patterns and structures in data without explicit guidance or labeled examples.

Koutelida Xristina

2024-04-22 13:59:21 UTC

Imagine teaching a robot to navigate a maze. At first, it moves randomly, but when it reaches the end, you give it a reward (like a treat). Over time, the robot learns to associate certain actions (like turning left or right) with positive outcomes (reaching the end of the maze) and adjusts its behavior accordingly. Reinforcement learning involves training an AI agent through trial and error to maximize cumulative rewards in a given environment.

Imagine your child starting a new school and having to come back at home, alone, by bus. For the first trip, you go with your kid and show him all the stops (name, looks) until he could reach the stop that takes him home. The other day he will be able to come back home by himself.

2024-04-22 16:22:36 UTC

Imagine a child tidying up his toys and sorting them on different boxes. At some point, he finds a toy that does not belong to him. It has been forgotten by a guest. The child will recognize it's not his and he runs to his parents to show it.

irynna_olteanu — 2024-04-22 16:30:27 UTC

Imagine a child playing with toy with different shapes. He needs to put different shapes (triangle, circle etc.) on their place. The child will try to fit in the shapes by trying, mistaking until he finaly reaches the right ones.

irynna_olteanu — 2024-04-22 16:33:45 UTC

enriccervello — 2024-04-24 11:10:29 UTC

Supervised ML AI is like giving a set of simple exercises to your students, that can be solved with a simple operation and you already show the answers. They can easily check their answers.

enriccervello — 2024-04-24 11:12:24 UTC

Unsupervised ML AI is like giving a set of exercises to your students, that cannot be solved with a simple operation, and giving them only the final answer, hoping they are able to find it.

enriccervello — 2024-04-24 11:13:39 UTC

Reinforced learning is like controlling each step of an exercise, in order to assure that the way of finding the right answer is correct.

Igors

igorjok17lv — 2024-04-24 19:02:00 UTC

During a history lesson teacher use gamification method. Ask student imagine playing a video game where they control a character that needs to navigate through a maze to find a treasure. Every time student character makes a move that gets closer to the treasure, the game gives you points. If student character hits a wall or goes the wrong way, they might lose points or have to start over. As student play more, they learn which paths lead to the treasure and which ones don’t, just by trying different routes and seeing what happens. Student learn the best actions to take based on rewards (like gaining points) or penalties (like losing points).

Igors

igorjok17lv — 2024-04-24 19:04:50 UTC

At the history lesson task to students. Students need imagine they are an archaeologist who has just discovered a hidden room full of various unsorted artifacts. No one has labeled them, and it's students job to sort them into groups that seem to go together, maybe by material, use, or age. By looking for patterns and similarities, students create own categories.

Igors

igorjok17lv — 2024-04-24 19:05:53 UTC

At the history lesson students identify different historical artifacts. Teacher shows to students pictures of ancient tools, weapons, and pottery, and tells them which civilization each one came from. As student see more examples, them get better at figuring out the origins of new artifacts on your own.

Eva

toth_eva — 2024-04-25 10:56:36 UTC

This is where most fabrications occur?

Eva

toth_eva — 2024-04-25 10:57:30 UTC

Sandra Magnuszewska

2024-04-28 10:56:11 UTC

A situaction at the gym: a coach shows you different types of exercises and gives names of them. You try to remember them.

Sandra Magnuszewska

2024-04-28 10:59:02 UTC

You plant roses. They grow and they are of different colours, however suddenly some kind of weed pop up. It does not belong to this group and we usually remove it as it is 'different'

Sandra Magnuszewska

2024-04-28 11:00:41 UTC

You bake a cake but you don't know the weight and the ration of ingredients to make it and finish successfully. You make many attempts to make it good.

Cira Serio - Italy

2024-05-01 11:36:55 UTC

It's not an easy task, but I like challenges and I also think that only by putting ourselves out there can we become better teachers for our students. I have a class 4 and a class 5 that I am already working with to help them understand how AI can help us on our learning journey and at the same time we also stop to reflect on the possible disadvantages that incorrect or excessive use of AI could cause in our lives. I watched with my students the interesting film proposed in this course that I found really interesting.

I asked my students to imagine that we are teaching a robot to recognize colors. We show it different images of objects (such as apples, bananas, and tennis balls) and tell the robot what color each object has. The robot learns to connect colors to different shapes and objects. Now, when we show it a new image, the robot can guess the color of the object based on what it has previously learned.

Cira Serio - Italy

2024-05-01 11:40:38 UTC

I asked my students to imagine that they had a lot of bricks of different colors and shapes. We want to organize them into similar groups without explicitly telling them how to do it. Unsupervised learning helps us find patterns or similarities between bricks. The robot could group the red bricks together and the blue ones into another group, without knowing in advance what colors are there.

Furthermore, Tutor students helped children in the younger classes learn about how AI learns using the game made available by the Code.org platform

Cira Serio - Italy

2024-05-01 11:49:02 UTC

I asked to think of a game in which a virtual character has to overcome obstacles to reach a treasure. Initially, the character doesn't know how to overcome obstacles. Reinforcement learning works like a virtual trainer: every time the character makes a correct move (e.g., jumps over a ditch), they receive a "reward" (points). The character learns to make better decisions in the game to maximize points and reach the final treasure.

Mariana Borzei

marianaborzei — 2024-05-01 14:03:57 UTC

Facial recognition in photographs or videos.

Mariana Borzei

marianaborzei — 2024-05-01 14:09:13 UTC

Rule association to discove relationships between items in datasets, such as product recommendations based or purchase history.

Mariana Borzei

marianaborzei — 2024-05-01 14:13:19 UTC

Models for autonomous driving are updated in real-time with data on traffic conditionsand vehicle behavior to ensure safe and efficient driving.

Süheyla,Türkiye

syuksel47 — 2024-05-02 06:09:06 UTC

I have him draw a picture of his favorite hero. I show him how to animate it. He does the animation himself.

Süheyla,Türkiye

syuksel47 — 2024-05-02 06:12:32 UTC

I have him draw a picture of his favorite hero. I show him how to animate him. He does the animation himself. The character is animated in different ways.

Süheyla,Türkiye

syuksel47 — 2024-05-02 06:13:37 UTC

I have him draw a picture of his favorite hero. I show him how to animate it. He does the animation himself. The character is animated in different ways. The students do the animation themselves.

Pompilia Cosma

CosmaPompilia — 2024-05-05 05:56:04 UTC

I teach students with special needs

Supervised learning is like teaching a child with examples. Imagine a teacher showing pictures of fruits and telling the child what each fruit is called. The child learns by seeing many examples and their correct labels. Then, when shown a new fruit, the child can guess its name based on similarities to the examples they've seen before.

Pompilia Cosma

CosmaPompilia — 2024-05-05 05:56:34 UTC

I teach students with special needs

Unsupervised learning is like sorting toys into different groups without any labels. Imagine you have a big box of toys and you want to organize them. Without anyone telling you which toys belong together, you start to group them based on similarities like color, shape, or size. Over time, you might notice patterns and create different groups without being told what category each toy belongs to.

Pompilia Cosma

CosmaPompilia — 2024-05-05 05:57:06 UTC

I teach students with special needs

Reinforcement learning is like teaching a dog tricks with treats. When you're teaching a dog a new trick, you give them treats when they do something right and ignore them when they do something wrong. The dog learns to associate certain actions with rewards and will repeat the actions that earn treats. Similarly, in reinforcement learning, an AI learns by receiving rewards or punishments based on its actions, helping it improve its decision-making over time.

Cristina-Georgiana Voicu

erasmusplusproject20232024_ — 2024-05-09 05:23:36 UTC

Supervised Machine Learning (ML) AI:
- Explanation: Supervised machine learning involves training a model on a labeled dataset, where each input is associated with a corresponding target label. The model learns to map inputs to outputs based on the provided examples.
- Analogy: Think of a teacher guiding a student through a set of practice exercises with answers provided. The teacher corrects the student's mistakes and helps them learn the correct associations between questions and answers. The student eventually learns to solve similar problems on their own.

Cristina-Georgiana Voicu

erasmusplusproject20232024_ — 2024-05-09 05:23:54 UTC

Unsupervised Machine Learning (ML) AI:
- Explanation: Unsupervised machine learning involves training a model on an unlabeled dataset, where the algorithm tries to find patterns or structures in the data without explicit guidance.
- Analogy: Imagine organizing a messy room without any instructions on where things should go. You start categorizing items based on similarities and differences, grouping similar items together. Through this process, you might discover hidden patterns or relationships among the items, helping you organize them more efficiently.

Cristina-Georgiana Voicu

erasmusplusproject20232024_ — 2024-05-09 05:24:13 UTC

Reinforcement Learning:
- Explanation: Reinforcement learning is a type of machine learning where an agent learns to make decisions by interacting with an environment. The agent receives feedback in the form of rewards or penalties based on its actions, helping it learn the optimal strategy to achieve its goals.
- Analogy: Consider teaching a dog new tricks using treats as rewards. You give the dog a command, such as "sit" or "roll over," and reward it with a treat when it performs the correct action. Over time, the dog learns to associate specific actions with rewards and adjusts its behavior to maximize the treats it receives, effectively learning the desired tricks.

Eugenia Nour

2024-05-09 14:31:10 UTC

Invatarea automata presupune un set de algorimi care se repeta si este adaptata unor sarcini de lucru si care pot verifica de fiecare data rezolvarea corecta.

2026-07-27 13:40:12 UTC

Supervised Learning (Apprendimento Supervisionato):
È come un insegnante di scuola guida che ti corregge durante i quiz. Tu rispondi a tante domande e, per ogni errore, l'istruttore ti dice subito se la risposta è giusta o sbagliata, finché non impari a riconoscere tutti i cartelli stradali. L'algoritmo riceve dati etichettati dall'uomo (es. foto con l'etichetta "questo è cyberbullismo" o "questo è un messaggio sicuro") e impara a classificarli.
Unsupervised Learning (Apprendimento Non Supervisionato):
È come dividere una pila enorme di vestiti appena lavati senza etichette o istruzioni: nessuno ti dice cosa sono, ma tu inizi a metterli in mucchi diversi basandoti sulle somiglianze (le felpe con le felpe, i pantaloni con i pantaloni, i colori scuri separati dai chiari). L'algoritmo esplora dati non etichettati per trovare autonomamente strutture nascoste, pattern o gruppi (cluster).
Reinforcement Learning (Apprendimento per Rinforzo):
È come imparare a giocare a un videogioco (es. Super Mario) provando e riprovando. All'inizio muovi il personaggio a caso: se cadi in un dirupo ricevi una penalità (nessun punto), se salti sul fungo guadagni punti. A tentativi ed errori, il personaggio impara qual è la strategia migliore per finire il livello. L'agente impara compiendo azioni in un ambiente e ricevendo premi o punizioni in base ai risultati.