AI and Other Technologies

Transcript of the free preview lesson. Remaining lessons unlock with the full course.

Welcome to AI and other technologies. In these lessons, you'll explore AI and machine learning and discuss how you can use AI to design robotics. After completing these lessons, you'll know how to combine data analytics with artificial intelligence. Machine learning is one of the branches of artificial intelligence and it allows a machine to learn from training data sets and bases its learning on past performance. There are two types of machine learning, unsupervised and supervised algorithms. Unsupervised algorithms provide conclusions from real life data sets. The input data isn't labeled like a supervised algorithm. The main use case is for analysis purposes and it doesn't have a workaround for passing feedback to the actual algorithm. Two examples are K means and hierarchical algorithm. Unsupervised learning algorithms don't use previous learning. The data is known as unlabeled data and is passed through the machine to generate the results. If you're required to identify fraudulent transactions, it can be handled using unsupervised learning. You can create a machine learning program that can analyze the data and generate a list of transactions. It can easily categorize them into fraudulent and non-fraudulent transactions. A supervised algorithm uses training data sets and allows us to validate the model. Once the model is tested, then it will work with new data sets. The input data is often labeled in this case and the algorithm is tested on real data sets and the feedback applied to the program. The program is also modified based on the training data sets. The most important use cases are prediction use cases. Here are a few examples related to the algorithm. Decision tree, logistic regression and support vector machine. Consider the use case that can be solved using this model. You're monitoring turbine performance and need to generate alerts whenever the turbine trips. The operator is currently doing this work by analyzing the data and predicting failure based on certain data points, you'll create a supervised algorithm that can easily learn from this model and you know the expected output. Your team has previous experience in solving similar problems. This type of problem solving is known as classification. Your machine learning algorithm already knows how to handle the situation from training data and once you publicize the program in the actual production environment, it can perform the same task that the operator is doing manually. You'll notice that the machine learning model uses test data and provides predictions. Once the machine learning algorithm is ready for production, then you can use the machine learning model. The machine will accept the labeled data as the input and pass it through the machine learning model. The model generates the outcome based on previous learning. Supervised Learning uses these steps. 1. Extract historical data. 2. Generate sample data. 3. Create the training data sets. 4. Create the machine learning algorithm. 5. Pass the training data set into a machine learning algorithm. 6. build a statistical model. 7. perform validation and prediction. And 8. Generate the model outcome. Supervised learning follows the model where data and the response are known. You'll create the model based on this input. It's also considered a complex method for solving a problem, and it requires many resources and needs time to compute. Unsupervised learning allows developers to create a model and pass the input data. The predicted response is generated from model and input data and is considered simple compared to the unsupervised algorithm.