WEBVTT 00:00.080 --> 00:05.640 Let's learn about the third type of machine learning, which is unsupervised learning. 00:05.880 --> 00:14.800 Unsupervised learning is a type of machine learning that analyzes and models data without labeled responses 00:14.960 --> 00:17.640 or predefined categories. 00:17.840 --> 00:25.880 Unlike supervised learning, where the algorithm learns from input output pairs, the unsupervised learning 00:25.880 --> 00:36.000 algorithms work solely with input data and aim to discover hidden patterns, structures, or relationships 00:36.000 --> 00:43.880 within the data set independently, without any human intervention or prior knowledge of that data. 00:43.880 --> 00:46.080 Meaning, it's very clear here. 00:46.080 --> 00:48.160 We have training data set. 00:48.200 --> 00:54.760 We have training data set with no labels passed by the algorithm creating the model. 00:54.760 --> 01:01.920 And then we get the prediction under supervised learning for the clustering dimension reduction and 01:01.920 --> 01:03.760 association rule mining. 01:03.920 --> 01:10.360 Clustering A clustering technique involves grouping similar data based on defined criteria. 01:10.560 --> 01:15.660 It's useful for segmenting data and finding patterns in each group. 01:15.780 --> 01:17.220 Dimension reduction. 01:17.220 --> 01:23.740 In order to find the exact information, dimension reduction reduces the number of variables considered 01:24.020 --> 01:25.900 and the association rule. 01:25.940 --> 01:34.420 Mining discovering relationships between seemingly independent databases or other data repositories 01:34.420 --> 01:36.500 through association rules. 01:36.540 --> 01:45.460 I prepared a very good example about unsupervised learning, and this image shows sets of animals like 01:45.460 --> 01:53.420 elephants, camels and cows that represent raw data that that the unsupervised learning algorithm will 01:53.420 --> 01:54.300 process. 01:54.460 --> 01:57.060 Here we have the input raw data. 01:57.100 --> 02:05.940 The interpretation stage signifies that the algorithm doesn't have predefined labels or categories for 02:05.940 --> 02:06.620 that data. 02:06.740 --> 02:14.500 It needs to figure out how to group or organize the data based on the inheritance pattern. 02:14.660 --> 02:23.800 An algorithm represents an algorithm represents unsupervised learning process, which can be clustering, 02:23.800 --> 02:29.440 dimensionality reduction, or anomaly detection to identify patterns in the data. 02:29.480 --> 02:33.400 So based on this algorithm, the model will be trained. 02:33.640 --> 02:38.520 The process stage shows the algorithm working on the data. 02:38.800 --> 02:43.120 The output shows the result of unsupervised learning process. 02:43.120 --> 02:50.200 In this case, the algorithm might have grouped the animals into clusters based on their species, for 02:50.200 --> 02:55.160 example elephants, camels, cows, and the output. 02:55.200 --> 03:03.520 Here, for example, you can count the number of legs and number of eyes that all of of necks. 03:03.680 --> 03:11.520 So those are some variables that may be used for processing and may be used by the algorithm in order 03:11.520 --> 03:13.080 to process the data. 03:13.360 --> 03:18.920 The working of unsupervised machine learning can be explained in many steps. 03:18.920 --> 03:23.960 The first step is collecting the data like the input raw data. 03:23.960 --> 03:30.160 Gather data set without predefined labels or categories without any tags exactly like this image. 03:30.200 --> 03:33.260 Go and get those images. 03:33.260 --> 03:39.300 We don't tell the machine that those four cows or camels or elephant. 03:39.300 --> 03:41.220 Select the algorithm. 03:41.340 --> 03:46.220 Choose a suitable supervised unsupervised algorithm such as clustering. 03:46.260 --> 03:47.660 Like k means. 03:47.860 --> 03:51.140 This is a very popular algorithm. 03:51.300 --> 03:58.300 K mean association rule learning like dimensionality reduction like PCA based on the goal. 03:58.340 --> 04:06.780 Also we have PCA train the model on raw data, feed the unlabeled data set to the algorithm. 04:06.780 --> 04:12.100 We get the data and all those data to the algorithm. 04:12.140 --> 04:19.860 The algorithm looks for similarities, relationships or hidden structure within the data, then grouping 04:20.140 --> 04:21.780 or transform data. 04:21.980 --> 04:30.980 The algorithm organizes data into groups and clusters, analyzing that discovered groups, rules or 04:30.980 --> 04:39.820 features to gain insight or use them for further tasks like visualization, anomaly detection, or as 04:39.820 --> 04:41.820 an input for other models.