You will find an extensive amount of examples of the use of machine learning related problems in daily life, since they solve the difficult parts of the available problems that are widely used techniques or algorithms. We often use many desktop or web-based applications that solve your problems out of the data even without knowing that what underlying techniques have been used. You will be wondered to know that many of them actually use widely used machine learning algorithms to make your life easier. There are many machine learning problems around. Here we will mention some example problems that really represent what machine learning is all about:
- Spam detection or spam filtering: Given some e-mails in an inbox, the task is to identify those e-mails that are spam and those that are non-spam (often called ham) e-mail messages. Now the challenging part is to develop an ML application that can be applied so that it can identify only the non-spam e-mails to stay in the inbox. and move the spam emails to the corresponding spam folder or delete them permanently from the email account. A typical example could be what you may do while using Gmail manually, but if you have an ML application, that application will do it automatically.
- Anomaly detection or outlier detection: The anomaly detection deals with the identification of items, events, or observations that are unexpected or non-confirming to the expected patterns in a dataset; in other words, the identification of suspect patterns. The most common example is network anomaly detection using some machine learning applications. Now the challenging task is to develop an ML application that can be applied successfully to simply identify the unusual data points from the data propagating across the network.
- Credit card fraud detection: Credit card fraud is very common nowadays. Stealing credit card related information from online shopping and using it in an illegal way happens in many countries. Suppose you have a transactional database for a customer for a particular month. Now the challenging task is to develop an ML application to identify those transactions that were made by the customer themselves and those done by others illegally.
- Voice recognition: Recognizing a voice and converting it into a corresponding text command and later performing some actions, as an intelligent agent does. The most widely used applications include Apple Siri, Samsung S-Voice, Amazon's Echo (consumer space), and Microsoft Cortana (especially because Cortana has SDKs for extensibility and integration, and so on). Another example would be locking or unlocking your smartphone by using the recognised voice.
- Digit/character recognition: Suppose you have a handwritten zip code or address or message on/inside an envelope, now the task of digit/character recognition is to identify and classify the digits or characters for each handwritten character that is made by different people. An efficient ML application could help in this regard to read and understand handwritten zip codes or characters and sort the contents of the envelope by the geographic region, or more technically, by the image segmentations.
- Internet of Things: Large-scale sensor data analytics for prediction and classification from real-time streamed data. For example, smart living room monitoring including water level checking, room temperature checking, home appliances controlling, and so on.
- Gaming analytics: Analytics for sports, games, and console-based gaming profiles in order to predict upsell and target in-app purchases and modifications.
- Face detection: Given a digital photo album of hundreds or thousands of photographs, the task is to identify those photos that resemble a given person. An efficient ML application, in this case, could help to organise photos by person.
- Product recommendation: Provided a purchase history of a customer along with a large inventory of products, the target is to identify those products that the customer will likely be interested in purchasing with an ML system. Business and tech giants such as Amazon, Facebook, and Google Plus have this recommended feature for the users.
- Stock trading: Given the current and historical prices for a stock market, predict whether stock should be bought or sold in order to profit with the help of an ML system.
The following are some examples of machine learning that are emerging and the demands of current research:
- Privacy preserving data mining: Mining customer's purchase rules from the maximal frequent pattern and association rules from business oriented retail databases to increase purchases in the future
- Author name disambiguation: Disambiguation performance is evaluated with manual verification of random samples of pairs from clustering results from a list of authors from a set of given publications
- Recommendation systems: Recommender system based on click stream data using association rule mining
- Text mining: Plagiarism checking from a given text corpus for example
- Sentiment analysis: A lot of decisions these days are being made by business and tech companies based on the opinion of others, and it will be a good place to innovate machine learning
- Speech understanding: Given an utterance from a user, the target is to identify the specific request made by the user. A model of this problem would allow a program to understand and make an attempt to fulfill that request. For example, iPhone with Siri and Samsung Voice Recorder in meeting mode have this feature implemented
Some of these problems are the hardest problems in artificial intelligence, natural language processing, and computer vision that can be addressed and solved using ML algorithms. Similarly, we will try to develop some ML applications emphasizing these problems in upcoming chapters.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.