Machine learning in the cloud

Summary

The video script explores the challenges of teaching computers to understand the complex, messy world as humans do. It discusses the evolution of artificial intelligence, machine learning, and deep learning, emphasizing their applications in everyday life like Google Maps and search recommendations. The script highlights the importance of training data in machine learning, the adaptability of algorithms across different use cases, and the ease of entry into the field due to advancements in technology. It also introduces Google Cloud's machine learning options, including BigQuery ML, AutoML, custom training, and pre-built APIs.

Takeaways

• 🌐 The world is complex for computers, which lack the intuitive understanding humans have for everyday objects.
• 💡 Early computing relied on logic, but real-world challenges require systems that can learn from their environment.
• 🤖 Artificial Intelligence (AI) aims to create machines capable of learning from their surroundings and improving over time.
• 📈 There are various approaches to AI, including pattern recognition, neural networks, reinforcement learning, and statistical inference.
• 🔄 Progress in AI is tied to advancements in technology, such as increased computing power and larger data sets.
• 🌟 AI applications are ubiquitous, often rebranded once they become commonplace, like search recommendations and language translation.
• 📊 Machine Learning (ML) is a subset of AI that uses data examples to enable machines to learn and improve without explicit programming.
• 📚 Training an ML model requires a dataset with inputs and corresponding labels to teach the model to recognize patterns.
• 🔍 Deep Learning is a subset of ML that adds layers for more complex learning, particularly useful for unstructured data like images and speech.
• 🔧 Algorithms or ML models are standardized and can be applied to different use cases after being trained on specific data.
• 🌉 Google Cloud provides multiple options for building ML models, including BigQuery ML, AutoML, custom training, and pre-built APIs.

Q & A

What is the difference between machine learning and traditional computing logic?

Traditional computing relies on hard logical rules, while machine learning allows computers to learn from data, examples, and mistakes, rather than being explicitly programmed for every possible scenario.

How has machine learning improved the things we use daily, like Maps or search engines?

Machine learning enhances tools like Maps and search engines by allowing them to learn from vast amounts of data. This enables features such as real-time traffic updates, more accurate search results, and personalized recommendations.

What is the significance of data in machine learning models?

Data is crucial in machine learning because it helps train the models. A machine learning model is only as good as the data it learns from, requiring large amounts of high-quality, labeled examples to make accurate predictions.

How does deep learning differ from other forms of machine learning?

Deep learning is a subset of machine learning that uses neural networks with multiple layers, enabling it to work with unstructured data like images, speech, or natural language, while other machine learning methods often handle structured data.

What role does artificial intelligence (AI) play in modern technology?

AI allows machines to mimic human intelligence and make decisions based on patterns in data. It powers a wide range of technologies, from anti-lock braking systems to email spam filters, and translation services.

How do neural networks help machines recognize images or objects?

Neural networks learn to recognize images by processing input through multiple layers, starting with simple patterns like edges and colors, and gradually building towards identifying objects such as cats or dogs.

What are some common methods used in AI and machine learning today?

Common methods include pattern recognition, artificial neural networks, reinforcement learning, statistical inference, supervised learning, and unsupervised learning. Each method serves different purposes and can be combined for better results.

Why has machine learning made significant progress in recent years?

Machine learning has advanced due to improvements in technology, including faster computers, larger datasets, and better algorithms, allowing for more complex models and solutions at scale.

What are some tools Google Cloud provides for building machine learning models?

Google Cloud offers tools like BigQuery ML for SQL-based models, AutoML for no-code solutions, custom training for coding flexibility, and pre-built APIs for using pre-trained machine learning models.

Why don’t machine learning models use logical rules like traditional programming?

Machine learning models rely on patterns and data to learn and make decisions. They use functions, not logical rules, to differentiate between categories in data, such as images, making them adaptable to different problems.

Outlines

00:00

🌐 Understanding the Messy World Through Machine Learning

This paragraph introduces the challenges computers face in understanding the complex and nuanced world around us, which humans can easily interpret. It discusses how machine learning is being used to improve various technologies such as Google Maps, search recommendations, video suggestions, and translations. The video 'Making Sense of a Messy World' is referenced, where Google engineers and researchers explain the evolution of computing from logic-based to machine learning-based approaches. The narrative emphasizes the shift from hard logical rules to learning systems that can adapt and improve over time, touching on various subfields like pattern recognition, artificial neural networks, reinforcement learning, and probabilistic machine learning. The importance of computational power and large data sets for advancing AI is highlighted, along with the idea that intelligence in machines is not a singular achievement but a collection of capabilities working together.

05:02

🤖 The Evolution of Machine Learning and Deep Learning

This section delves into the definitions and differences between artificial intelligence (AI), machine learning, and deep learning. AI is presented as an overarching term for technologies that mimic human intelligence, while machine learning is described as a subset that uses data examples to enable computers to learn without explicit programming. Deep learning is introduced as a subset of machine learning that adds layers for more profound learning, especially useful for unstructured data like images and speech. The paragraph explains how machine learning models are trained using examples, with inputs and corresponding labels, and how the quality of the training data directly impacts the model's performance. It also touches on the standardization of machine learning algorithms, which can be applied to various use cases after being trained on specific data sets. The excitement around machine learning is attributed to the lowered barriers to entry due to increased data availability, algorithm sophistication, and computing power.

10:04

🎨 Interactive Learning with Neural Networks

The final paragraph focuses on practical applications of machine learning, specifically mentioning Google Cloud's machine learning options. It outlines four approaches: BigQuery ML for SQL-based machine learning, AutoML for no-code model creation, custom training for tailored machine learning environments, and pre-built APIs for utilizing existing models. The paragraph invites viewers to participate in a doodling game at quickdraw.withgoogle.com, which contributes to a doodling data set for machine learning research. It also introduces deep neural networks (DNNs) and how they mimic the human brain's perception process, learning from basic visual elements to complex decisions. The TensorFlow Neural Network Playground is mentioned as a resource for understanding how these models work, highlighting the progress in machine learning that allows for scalable model building, exemplified by Google Photos.

Keywords

💡Artificial Intelligence (AI)

Artificial Intelligence (AI) is an umbrella term that encompasses various technologies and algorithms designed to mimic human intelligence in machines. In the video, AI is discussed as a field that has evolved from simple logic-based systems to complex ones that can learn from their environment. It is used to describe systems that can perform tasks such as recognizing images, translating languages, and recommending content.

💡Machine Learning

Machine Learning is a subset of AI that focuses on developing algorithms that enable computers to learn from data. In the script, it is highlighted as a tool set that allows computers to solve problems at scale without being explicitly programmed to perform the task. Machine learning is portrayed as a key technology that helps computers understand and react to the messy, unstructured nature of the real world.

💡Deep Learning

Deep Learning is a subset of machine learning that involves artificial neural networks with several layers, allowing the model to learn from data at a deeper level. The video script mentions deep learning in the context of handling unstructured data like images, speech, and text. It is a technique that enables computers to classify and understand complex patterns within data.

💡Pattern Recognition

Pattern Recognition is a subfield of machine learning that focuses on identifying regularities and patterns within data. In the video, pattern recognition is alluded to as a method for teaching machines to recognize and categorize different types of objects or data points, such as distinguishing between a rain cloud and other types of clouds.

💡Neural Networks

Neural Networks are computing systems inspired by the biological neural networks that constitute animal brains. The script refers to artificial neural networks as a method for reinforcement learning, where the network learns to make decisions based on patterns it recognizes in input data, akin to how the human brain processes stimuli.

💡Data

Data is the raw material that machine learning models use to learn. The script emphasizes that the quality of a machine learning model is directly related to the quality and quantity of data it is trained on. Data provides the examples needed for models to make accurate predictions and classifications.

💡Training

Training in the context of machine learning refers to the process of teaching a model to make predictions or decisions based on a dataset. The script describes how models are trained with examples, each consisting of an input and a corresponding label, which is the correct answer for that input.

💡Algorithms

Algorithms are the step-by-step procedures that machine learning models follow to perform tasks. The video script mentions that machine learning algorithms are standardized and can be applied to different use cases after training. Algorithms are the foundation of machine learning, dictating how data is processed and analyzed.

💡BigQuery ML

BigQuery ML is a tool mentioned in the script that allows users to create and execute machine learning models using SQL queries within BigQuery. It is an example of how machine learning can be integrated into existing data platforms to enable scalable and accessible AI applications.

💡AutoML

AutoML, as discussed in the script, is a no-code solution that allows users to build machine learning models through a point-and-click interface. It democratizes machine learning by enabling users without extensive coding knowledge to create and train custom models.

💡Custom Training

Custom Training refers to the process of developing a machine learning model from scratch, which includes coding the training environment and deploying the model. The script positions custom training as an option for those who require flexibility and control over the machine learning pipeline.

💡Pre-built APIs

Pre-built APIs are application programming interfaces that provide access to pre-trained machine learning models. The video script suggests using pre-built APIs as a way for users to leverage AI capabilities without the need to build and train their own models, which can be beneficial for those lacking the necessary data or expertise.

Highlights

The world is often messy and complicated for computers.

Machine learning is improving computers' ability to understand the world.

Machine learning enhances applications like Maps, search, video recommendations, and translations.

Early computing relied on logic, but real-world challenges require learning systems.

Artificial intelligence aims to build machines that learn from their environment.

There is no single 'right path' in artificial intelligence; various approaches are being explored.

The progress in AI depends on advancements in technology and data availability.

Artificial intelligence is now integrated into everyday applications, often without us realizing.

Intelligence in machines requires training with examples.

Machine learning models learn from a dataset of inputs and their corresponding labels.

The quality of machine learning models is directly tied to the quality of the training data.

Machine learning algorithms can be applied to various use cases once trained.

Deep learning is a subset of machine learning that works with unstructured data.

Image classification is a common application of deep learning.

Machines become intelligent through training and do not start out intelligent.

Google Cloud offers various options for building machine learning models.

BigQuery ML allows for creating and executing machine learning models using SQL queries.

AutoML provides a no-code solution for building machine learning models.

Custom training offers flexibility and control over the machine learning pipeline.

Pre-built APIs allow the use of machine learning models already trained by Google.

The barriers to entry for machine learning have fallen due to increased data, mature algorithms, and computing power.

Deep neural networks mimic the human brain's perception of stimuli.

The TensorFlow Neural Network Playground is an interactive tool for understanding machine learning models.

Transcripts

the world is filled with things that

most of us can understand and react to

without much thought for example a stop

sign partially covered by snow is still

a stop sign and a chair that's five

times bigger than usual is still a place

to sit

but for computers which don't have the

benefit of growing up and learning the

nuances of these objects the world is

often messy and complicated

in this first section you'll start by

watching a video called making sense of

a messy world where Google engineers and

researchers discuss how machine learning

is improving computers and many of the

things we use them for such as Maps

search recommending videos and

translations

let's watch this short video now

yeah thank you

[Music]

there have been a number of shifts in

the way we think about computing

over the past few decades the

terminology artificial intelligence has

come in and out of favor in the

scientific Community sometimes it's

called machine learning we tend to call

it machine intelligence these days I

just call it intelligence and sometimes

it's just the effort to build machines

that are better

so in the early days everything was

built on logic

doing mathematical integration problems

playing chess

but we realized that what the real

challenges were were the things that

people can do every day the real world

is actually very messy

hard logical rules are not the way to

solve really interesting real world

problems you have to have a system

that'll learn to get the knowledge in

you can't just program it all in

artificial intelligence is an effort to

build machines that can learn from their

environment from mistakes and from

people we're still at the stage where we

don't know what is the right path and

the right breakthrough so I mean there's

certainly a whole raft of different

approaches one of the subfields we call

pattern recognition artificial neural

network reinforcement learning for

example statistical inference and

probabilistic machine learning

supervised learning unsupervised

learning and we're not quite sure what

technique is going to lead to better

systems and in fact it's probably not

one technique for everything it's

probably a bunch of different techniques

and combinations of those techniques any

progress we make in building truly

intelligent systems is going to depend

on progress in technology generally and

until recently we didn't have computers

that were fast enough or data sets that

were big enough to to do that and so

being able to take a particular problem

and spread it out over lots and lots of

machines is a very important approach

because it makes our research faster

so there's applications of artificial

intelligence around us all the time when

it begins to work or it does work it's

all of a sudden given another name we're

all already using it and very

comfortable with it things that now we

regard as routine

um 30 years ago would have been regarded

as amazing examples of artificial

intelligence anti-lock braking autopilot

systems for planes search

recommendations Maps

to decide whether or not this particular

email is Spam or not spam

the ability to translate one language to

another with your phone

ten years ago if you tried to talk to

your computer or to your phone you know

that would just be hopeless we we're

seeing a steady torrent of these tricks

one after the other getting figured out

right now and I think a lot of people

that are close to the field have this do

have that kind of breathless sense that

things are moving quickly it's a

progressive thing it's about building

things that are slightly better slightly

better slightly better intelligence is

really not going to be something that we

ever succeed in defining in a succinct

and singular way it's really this whole

constellation of capabilities that you

know all kind of are beautifully

orchestrated and working together

predicting the long-term future is very

difficult

nobody can really do it

and the bad thing to do is take

whatever's working best now

and assume the future is going to be

like that forever

[Music]

in that video you heard a few different

definitions of artificial intelligence

and machine learning let's explore the

differences together

artificial intelligence or AI is an

umbrella term that includes anything

related to computers mimicking human

intelligence

for example in an online word processor

robots perform human actions for

spelling and grammar checks

machine learning is a tool set like

Newton's Laws of mechanics

just as you can use Newton's laws to

learn how long it will take a ball to

fall to the ground if you drop it off a

cliff you can use machine learning to

solve certain kinds of problems at scale

by using data examples but without the

need for custom code

you might have also heard the term deep

learning or deep neural networks

deep learning is a subset of machine

learning that adds layers in between

input data and output results to make a

machine learn at more depth it's a type

of machine learning that works even when

the data is unstructured like images

speech video natural language text and

so on

image classification is a type of deep

learning a machine can learn how to

classify images into categories when

it's shown lots of examples

the basic difference between machine

learning and other techniques in AI is

that in machine learning machines learn

they don't start out intelligent they

become intelligent

so how do machines become intelligent

intelligence requires training

to train a machine learning model

examples are required

for example to train a model to estimate

how much you'll owe in taxes you must

show the model many many examples of tax

returns

or if you want to train a model to

estimate trip time between one location

and another you'll need to show it many

examples of previous Journeys

the first stage of ml is to train an ml

model with examples

an example consists of an input and the

correct answer for that input

this is called the label

in the case of structured data that's

rows and columns an input can simply be

a single row of data

in unstructured data like images an

input could be a single image of a cloud

that you want to classify as a rain

cloud or not

imagine you work for a manufacturing

company and you want to train a machine

learning model to detect defects in the

parts before they are assembled into

products

you'd start by creating a data set of

images of parts

some of those images would be good and

some parts would be defective

for each image you'll assign a

corresponding label and use that set of

examples to train the model

an important detail to emphasize is that

a machine learning model is only as good

as the data used to train it

and a good model requires a lot of

training data of historical examples of

rejected parts and parts in good

condition

with these elements you can train a

model to categorize Parts as defective

or not

the basic reason why ml models need high

quality data is because they don't have

human general knowledge data is the only

thing they have access to

after the model has been trained it can

be used to make predictions on data it's

never seen before

in this example the input for the

trained model is an image of a part

because the model has been trained on

specific examples of good and effective

Parts it can correctly predict that this

part is in good condition

algorithms or ml models are standard

that means that they exist independently

of the use case

although detecting manufacturing defects

in images and detecting disease leaves

and images are two different use cases

the same algorithm which is an image

classification network works for both

similarly standard algorithms predict

the future value of a Time series or to

transcribe human speech to text

resnet for example is a standard

algorithm for image classification

it's not essential to understand how an

image classification algorithm Works

only that it's the algorithm we should

use if we want to classify images of

automotive parts

when we use the same algorithm on

different data sets different features

or inputs are relevant to the different

use cases and we can see them

represented visually here

you might be wondering isn't the logic

different

you can't possibly use the same rules to

identify defects in manufacturing that

you use to identify leaves

the logic is different but machine

learning doesn't use logical rules

the image classification network isn't a

set of basic if this then that rules but

rather a function that learns how to

differentiate between categories of

images

so although you start with the same

standard algorithm after training the

trained model that classifies leaves is

different from the train model that

classifies parts

and you can actually reuse the same code

for other use cases that are focused on

the same kind of task

in our example you identified

manufacturing defects but the higher

level task classified images

this means you can reuse the same code

for another image classification problem

like finding examples of products in

photos posted on social media

however you still have to train it

separately for each use case

much of the excitement around ml is

because the barriers to entry have

fallen you don't need to be an

astrophysicist to do machine learning

this is because of the convergence of

several factors

the increasing availability of data

the increasing maturity and

sophistication of ml algorithms

and the increasing power and

availability of computing hardware and

software

Google Cloud offers four options for

building machine learning models

the first option is bigquery ml you'll

remember from the previous module of

this course that bigquery ml is a tool

for using SQL queries to create and

execute machine learning models in

bigquery if you already have your data

in bigquery and your problems fit the

predefined ml models this could be your

best choice

the second option is automl which is a

no code solution so you can build your

own machine learning models on vertex AI

through a point-and-click interface

the third option is custom Training

through which you can code your own

machine learning environment the

training and the deployment which

provides you with flexibility and

control over the ml pipeline

and finally there are pre-built apis

which are application programming

interfaces

this option lets you use machine

learning models that have already been

built and trained by Google so you don't

have to build your own machine learning

models if you don't have enough training

data or sufficient machine learning

expertise in-house

let's play a quick game where we'll see

how a neural network can learn how to

recognize doodling

during the process we'll help to teach

it by adding our drawings to the world's

largest doodling data set and help with

machine learning research

pause the video now and in a different

browser tab or window head to

quickdraw.withgoogle.com when you're

done head back here to continue the

video

here's the model behind the game we just

played it's called a deep neural network

or DNN

the layers are meant to mimic our own

human brains in the way we perceive

stimuli

with each layer the model learns more

about the image of this dog that's

hiding in a laundry basket starting from

basic edges and colors to a final

decision of cat or dog you'll build your

own image recognition model later in the

course spoiler alert you don't need to

write any code but if you want

additional information about how these

model types work see the tensorflow

neural network playground

the tensorflow neural network playground

at

playground.tensflow.org is a great

Interactive Learning tool for

understanding how computers think and

shows how far ml has come that we can

build these models at scale like in

Google photos