AI, machine learning and deep learning

Dr. Chou, Chun-Liang will explain the basic concept of AI this week. We will discuss the current clinical cases utilizing AI in pulmonary medicine and diagnosis.

In this video, he will start by interpreting the definition of artificial intelligence. AI can be simplified as the idea of building machines that are capable of thinking like humans such as reasoning, learning, language processing, and the display of knowledge of information. The basic mechanism is to use Big data to train the computer to build a model for predicting outcomes.

There are many branches of research in AI. Machine learning, natural language processing, expert systems, and vision are artificial intelligence’s main and important branches. Deep learning, predictive analysis, classifying, clustering, and image recognition are widely used in medicine, also in respiratory medicine.

Continue, Dr. Chou will use the chart to explain the differences between artificial intelligence, machine learning, and deep learning. They are often used interchangeably, but they are heterotic. Machine learning is a subset of AI using the component to recognize features, and patterns of input data and identify the relationship between data and the targets. Thus, deep learning learns a task automatically and improves from backpropagation with fine-tuning programming.

There are two major models given in deep learning: Supervised and non-supervised. In supervised learning, we give the data sets and already know what our correct data should look like, which means on the basis of data using labeling. Unlike supervised learning, we provide data sets with unknown targets and ask to find the patterns of clusters from the given data sets. Through this model, the computer discovers a pattern without any guidance.

Dr. Chou will use Immune profiling data in supervised learning as an example to explain more. Each input of the data set is matched to a specific output value. Unsupervised learning data sets are presented as clustering and the cocktail party. An algorithm is used to find the structures between the given datasets. By the algorithm, some unexpected clustering data will be explored. There are two clusters of immune cells identified by an unsupervised plot on the blood cells of patients with severe covenanting pneumonia. With the supervised protocol, these two clusters cannot be identified. While the unsupervised protocol is often used to identify patterns of unclassified or unlabeled data. Data points within each cluster are more similar than data points in other clusters.

Share and learn:

Could you tell the difference between supervised and unsupervised learning? Leave your thoughts in the comments section below.

Please also checked the download section where you could have the slides from the educator.