Recognizing text | Mediateca de EducaMadrid

As we already know, computers can only work with numbers, and yet, there are computer systems that 00:00:00

are capable of understanding our texts. How does this happen? What is used is a mechanism to 00:00:05

translate words or phrases into a numerical representation known as embeddings. As Jeremy 00:00:12

Howard mentions in his book AI Applications Without Having a PhD, the artificial intelligence 00:00:17

community sometimes likes to use somewhat pompous names for concepts that are actually very simple. 00:00:22

And this is somewhat the case with embeddings. 00:00:27

Let's see how they are built. 00:00:31

Let's imagine we are in this situation where a numerical representation has already been assigned to a set of words using two numbers. 00:00:34

Where would we place the word apple? 00:00:42

Near position A there are several round objects. 00:00:44

Near B there are words that have to do with constructions. 00:00:48

But in position C we would have the word apple near others related to fruits. 00:00:51

This would be a good location since the objective of embeddings is that similar words 00:00:55

correspond to nearby points and words that are different correspond to distant points. 00:01:01

Let's see another example. Suppose we have already assigned the numerical representation 00:01:07

to the words dog, puppy and calf. Where would we place the word cow? All three positions could make 00:01:12

some sense but if we place it in position C we would be capturing some relationships between 00:01:20

the words, which is precisely another one of the objectives of embeddings. 00:01:24

In this case we would be capturing two analogies. 00:01:29

On one hand, puppy is to dog, what calf is to cow. 00:01:33

And on the other, puppy is to calf, what dog is to cow. 00:01:40

Thus, this embedding would be capturing two properties of the words age and size. 00:01:43

And basically these are embeddings. 00:01:49

What happens is that the ones we use in real applications have hundreds or thousands of 00:01:52

that is to say, that a word translates to a vector of hundreds or thousands of numbers. 00:01:57

As we detail in the article associated with this video, these embeddings allow for visualizations 00:02:03

and classroom activities that are very interesting and that could be the 21st century equivalent of 00:02:08

learning to explore a dictionary. But these word embeddings have certain limitations when it comes 00:02:13

to recognizing sentences, since the same word can mean different things depending on the context. 00:02:18

Fortunately, since transformers were born with their attention mechanism that allows understanding 00:02:23

context, we now also have embeddings that are capable of assigning a numerical representation 00:02:29

to complete sentences in a coherent way. Thus, we can see that the sentence nothing 00:02:33

pleases me more than basketball is semantically closer to I love basketball than the sentence 00:02:45

I love football, despite the fact that these last two share more identical words. 00:02:50

And there are even multilingual sentence embeddings in which sentences that mean 00:02:59

the same thing in different languages receive a close numerical representation. 00:03:03

As we will see in upcoming episodes, these word and sentence embeddings are the foundation of 00:03:07

large language models like GPT-3 and Bloom. But while we get to that, don't stop playing 00:03:12

with the challenges and tasks we propose on our website, as they will allow you to interact 00:03:18

directly with the internal workings of many of the artificial intelligence systems we use daily. 00:03:23

Del mismo autor…

Programa con Scratch Jr y deja que desarrollen su imaginación programando.
Contenido educativo. subido por Felicisimo G. 00′ 59″ - hace 47 minutos - Idioma: - 4 visualizaciones
Proyecto Calendario con CANVA (Portada)
Contenido educativo. subido por Miguel G. 03′ 03″ - hace una hora - 4 visualizaciones
LA PÓCIMA DE OTOÑO
Contenido educativo. subido por Marta R. 06′ 29″ - hace una hora - 3 visualizaciones
SEMANA DIVERSIDAD 2025
Contenido educativo. subido por Laura C. 01′ 52″ - hace 2 horas - 5 visualizaciones
el plano de mi casa
subido por M. Belén D. 03′ 34″ - hace 9 horas - 2 visualizaciones
el plano de mi casa
subido por M. Belén D. 04′ 19″ - hace 10 horas - 2 visualizaciones
Five senses project
Contenido educativo. subido por Amaya C. 03′ 18″ - hace 10 horas - 4 visualizaciones