From the winners of the variety show "Jeopardy" and the masters of Go, to the disgraceful, ad-related racial characterization, we seem to have entered an era of rapid development of artificial intelligence. However, to create such a fully sensible person, his electronic "brain" can use the fair moral judgment to fully participate in complex cognitive tasks, and our current capabilities are not yet available.
Unfortunately, current developments have raised widespread fears about what artificial intelligence might become in the future. Its performance in recent popular culture shows how cautious and pessimistic we are about this technology. The problem with fear is that it can have serious consequences and sometimes encourage ignorance.
Understanding the internal workings of artificial intelligence is a good medicine to address these concerns. Moreover, this seriousness can lead to responsible and reassuring participation.
The core foundation of artificial intelligence is machine learning, an elegant and widely used tool. But to understand the meaning of machine learning, we first need to study how its potential absolutely exceeds its disadvantages.
Data is the key
Simply put, machine learning refers to teaching computers how to analyze data through algorithms to solve specific tasks. For example, for handwriting recognition, a classification algorithm can be used to distinguish letters written by different people. On the other hand, the housing dataset uses regression algorithms to estimate the selling price of a property in a quantifiable manner.
Then, machine learning ultimately comes down to data. Almost every business generates data in one way or another: Think about market research, social media, school surveys, and automation systems. Machine learning applications attempt to find hidden patterns and dependencies in the chaos of big data sets to develop models that predict behavior.
The data has two key feature samples and features. The former represents a single element in a group; the latter represents the features they share.
Take social media as an example: users are samples and their use can be translated into features. For example, Facebook uses different aspects of the "Like" activity (which vary from user to user) as an important feature for targeted advertising.
Facebook friends can also be used as a sample, and their connection with others can also be used as a feature to build a network that can study information dissemination.
My Facebook friends network: Each node is a friend who may or may not connect with other friends. The larger the node, the more connections there are. Similar colors also represent similar social circles.
In addition to social media, automated systems used as monitoring tools in industrial processes feature time snapshots of the entire process as samples, characterized by sensor measurements made at specific times. This allows the system to detect anomalies in the process in real time.
All of these different solutions rely on providing data to machines and teaching them to implement their own forecasts when strategically evaluating given information. This is machine learning.
Taking human intelligence as a starting point
Any data can be translated into these simple concepts, and any machine learning application, including artificial intelligence, uses these concepts as a basis for its construction.
Once the data is understood, it is time to decide how to process the information. One of the most common and intuitive applications of machine learning is classification. The system learned how to put data into different groups based on a reference data set.
This is directly related to the decisions we make every day, whether it's grouping similar products (such as kitchen items for beauty products) or choosing good movies based on past experience. Although these two examples may seem completely out of touch, they rely on a basic classification hypothesis: a prediction that is defined as a defined category.
For example, when we pick up a bottle of lotion, we use a specific list of features (such as the shape of the container, or the smell of the product) to accurately predict that it is a beauty product. A similar strategy is to predict whether a movie belongs to one of two categories by evaluating a set of characteristics (such as a director or an actor): good or bad.
By grasping the different relationships between the various features associated with a set of samples, we can predict whether a movie is worth watching, or, better yet, we can create a program to do this for us.
But to get this information, we need to be a data science expert, proficient in mathematics and statistics, and have enough programming skills to make Alan Turing and Margaret Hamilton proud. ? not completely.
In our daily lives, we have mastered enough native languages, even though only a few of us can get involved in linguistics and literature. The same is true for mathematics. It is always around us, so it is not a burden to change from buying or measuring raw materials to following recipes. Similarly, mastering machine learning is not a necessary condition for conscious and effective use of it.
Yes, there are really good and professional data scientists in the world, but anyone can learn the basics of data and improve the way they view and use information with little effort.
Solve problems through algorithms
Going back to the classification algorithm, let's consider an algorithm that mimics the way we make decisions. We are people in society, so what about social interaction? The first impression is very important. We all have an internal model to assess whether we like each other in the first few minutes of meeting others.
There are two possible outcomes: a good or bad impression. For each person, different characteristics (features) are taken into account (even unconscious) based on several encounters (samples) in the past. It may be tone or appearance, or courtesy.
For each new face we encounter, a model in our mind records these inputs and builds a prediction. We can decompose this model into a set of inputs, weighting them according to their relevance to the final result.
For some people, attraction may be very important, while for others, a sense of humor or a dog is more telling. Everyone will develop their own model, depending on her experience or data.
Different data leads to different models being trained and the results are different. Our brains develop mechanisms (although we don't fully understand this), but these mechanisms will determine how these factors will affect our weighting of factors.
What machine learning does is to develop precise and mathematical methods for the machine to calculate the results, especially if we can't easily handle the amount of data. Now more than ever, the data is huge and timeless. With a tool that can actively use this data to solve real-world problems, such as artificial intelligence, this means that everyone should and can explore and take advantage of this. We should do this so that we can not only create useful applications, but also put machine learning and artificial intelligence in a brighter, less worrying perspective.
There are a lot of resources for machine learning, but these resources do require some programming power. Many popular languages ​​for machine learning are available from basic tutorials to complete courses. In just one afternoon, you can start your adventure and get obvious results.
All of this is not to say that the concept of a machine with human thinking should not make us worry. But knowing more about how these ideas will work will enable us to be agents of positive change, allowing us to maintain control over artificial intelligence, not vice versa.
Electric Logistics Vehicle Battery Supply
Shenzhen Zhifu New Energy Co., Ltd. , https://www.sunbeambattery.com