AI and machine learning

1. AI and machine learning

Hi, and welcome back. Let's see what GCP has to offer in terms of artificial intelligence and machine learning.

2. The world of AI and ML

We start by defining what we mean by these terms. AI refers to the simulation of human intelligence in machines. These machines are designed to mimic human problem-solving and decision-making. Machine Learning, a branch of AI, is the process through which machines learn from data, and make decisions or predictions. This enables machines to improve their performance over time without being explicitly programmed for every task.

3. Data quality in ML

ML algorithms train themselves to perform a task. They train by ingesting training data and the quality of this training data is critical. The adage 'garbage in, garbage out' is particularly apt in the context of ML. High-quality data is clean, comprehensive, relevant, and up-to-date. When data meets these criteria, the predictions made by ML models are more accurate and reliable. Inaccurate data can lead to erroneous predictions, which could have a significantly adverse impact.

4. GCP's AI and ML Solutions

We already discussed the comprehensive suite of data tools GCP offers. Google’s global network ensures low latency and reliability, which are crucial for AI and ML applications. Building on top of this, GCP offers a comprehensive set of AI and ML solutions. The most basic of these services are accessible to users of any level of ML expertise.

5. Pre-trained models on GCP

GCP offers various pre-trained models available for out-of-the-box usage. Vision AI can label images, detect objects, and read handwritten documents. Video AI can help recognize places and objects as well as extract metadata.

6. Pre-trained models on GCP

The Natural Language AI helps extract insights from text. This could be extracting the topic of a long social media post. Or the sentiment of a customer review, whether it is positive or negative for instance. Of course, there is a lot more than can be done here. These pre-built models are readily available and can be easily integrated into applications. This feature enables businesses to leverage advanced AI capabilities without the large costs associated with model development and maintenance.

7. ML with BigQuery

BigQuery, GCP’s data warehousing solution, allows users to train basic ML models using BigQuery ML. This opens up model training to all users who are familiar with some analytics. The integration of analytics and ML enables a seamless transition from data analysis to the application of AI and ML, streamlining the process of extracting insights and value from data.

8. Vertex AI

While models can be trained in BigQuery, they are managed and deployed in Vertex AI. Vertex AI is an integrated machine learning platform designed to streamline the process of building, training, and deploying ML models. It brings together GCP’s services for ML on one platform, enabling data scientists and ML engineers to efficiently implement ML operations.

9. Vertex AI

On one hand, Vertex AI offers no-code ML solutions like AutoML. AutoML trains on the user's own data but automates complex model tuning tasks and offers an accessible pathway to leveraging AI. It is available for many different types of data including tabular, image, and text. For more advanced users, GCP offers custom training Vertex AI. Custom training on Vertex AI offers users more control and flexibility in building, training, and deploying machine learning models compared to AutoML. While AutoML automates much of the process, custom training allows for fine-tuning and customization, catering to advanced users with specific requirements and domain expertise.

10. AI in business

Businesses leverage GCP’s AI and ML solutions in numerous ways to create value. For example, in retail, a business can use BigQuery ML to train an ML model to predict customer churn. This model can be deployed on Vertex AI.

11. AI in business

A healthcare imaging service can use an ML model trained through AutoML on images labeled "Disease" or "Healthy". The trained model can label new images and give a provisional diagnosis for further workup.

12. AI in business

An online stock broker can employ custom training on Vertex AI. The recommendation model is trained on diverse sources of data based on a customer's profile, preferences, as well as stock price trends. Each model can be fine-tuned to the needs of the customer.

13. Let's practice!

Now let’s head over to the exercises and review what we have covered.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.