The SYPWAI platform was established to participate in the development of artificial intelligence. The service brings together representatives of companies that need complex monitoring via neural networks and experts who implement the necessary tools. Working with neural networks requires the concerted efforts of narrow-profile specialists and a dedicated platform for their training. SYPWAI helps to solve all the above mentioned problems with minimal investment of time and money
We managed to meet one of the key people of the SYPWAI project – Abdicant Derive
You once said that your team could improve the accuracy of neural networks. Can you share how you did it?
Every AI project requires two parts. First, properly labeled information. Usually marking is very simple, but this kind of work is monotonous and time consuming.
Second, the neural network itself must be written. Indeed, experts are creating a simulation of a small piece of human brain that can solve a specific problem, such as a living person. When a neural network learns, it is “rewarded” for a correct answer and “punished” for a wrong answer.
This simplifies the development of any solution. People who want to contribute will do their part. Experts check the results and create APIs for the obtained datasets.
What is the main idea of your project?
The following principle has become the main idea of the SYPWAI platform: anyone can participate in the formation of neural networks. The process is very similar to educational games for toddlers. The child sees a picture of a cat and answers which animal it is: a cat, a fox or a dog. The main job of our AI was to make predictions on how best to achieve this. It uses a list of options for different action sequences.
From an ideological point of view, the most important thing for SYPWAI is:
The effectiveness of neural network or social function training, getting people with disabilities to work, etc?
Our company has been working on artificial intelligence for a long time. In the process, we discovered that there is a major problem in this area: the company cannot allocate specific departments to data science that can create solutions using neural networks. Marking is a long process. At the same time, there are not many players and big companies in the AI market yet, as it is only forming. So, working with a neural network, we created a new profession: data annotation. Therefore, it is difficult to determine which of the components is more important.
How would you describe the work process at SYPWAI?
First, the person is given a test task. It is always very simple, so only people who have 99% correct answers can work. At this point the system will symbolically load the tokens, there is one percent cent. So the work began.
When most markers have the same answers, the information is transferred to the neural network. We believe that if seven out of ten choose one option and only three choose the other, the number will be exceeded. If they start answering wrong, the points will decrease and then the reward amount.
Let’s talk about business customers. What are their requirements?
We receive inquiries from companies, scientific and public organizations that need AI help. However, we are also developing requirements to solve the global problems humanity is constantly facing. Customers share their goals and concerns.
For example, a company develops prostheses. That’s why we help him find a solution so that their production is cheaper and the products of higher quality. Another example is a company interested in reducing crime in an area: in this case, AI will inform the most effective actions.
We interview customers and, based on the information obtained, train the direction of scientific research and the questions needed to learn AI to work with a particular company. The research helps to determine the right direction for the development of artificial intelligence and the answers to the questions help to define the needs more precisely.
Can you tell us a little more about the algorithms in your system?
From each order we receive a large amount of information: the company’s problems and tasks, goals set, available resources, existing scientific research related to the desired topic, answers from real people to surveys that help the company.
