Explainability in AI Systems 

Most of the AI system used in industries are unable to explain "why" a particular answer is proposed by the system. In many cases, "explainability" of the systems is crucial. For large number of situations, management may not be comfortable to make decision follow what AI propose simply because of the system say so. Evidences or supporting/similar experiences have to be shown to the management to support the claim or proposed solutions.

In one of our experiences, unlike traditional scoring system, loan approval officers are much more comfortable with a system's proposed solution, if the system can also provide reason why it propose to approve/reject a loan application. The supporting reason in this case can be statistics of historical cases previously occurred. By combining a machine learning called "Case-Based Reasoning" together with statistical methods, the historical cases are retrieved based on similar cases that has been found. This result in an explainable AI system that can propose its suggestion with the supporting reason.

i4.jpg

Ditto: Hybrid AI Engine

Best machine learning algorithm?

We are AI researchers who have been doing research on machine learning and AI related issues for more than 20 years. There is no such thing as "the best approach" when comes to building an AI system. There are hundreds of machine learning algorithms that can be used in building  systems, every single one of them has its own strengths and weaknesses. Even in the same domain, small changes occurred with the data, can easily lead to inaccurate result proposed by the AI systems. So, why not we use multiple approaches in building a system.

 

We do not believe in a single machine learning algorithm

Ditto, our hybrid AI engine, is a highly customizable AI engine that contains more than 10 machine learning algorithms to choose from. Unlike other tools for AI systems development, our engine can also be customized to run multiple algorithms in order to automatically find the best answer for you.

i1.png

Customized AI System

All problems are unique

No such thing as using a single approach as a silver bullet to solve all kinds of problems of customers. In fact, we do not believe that such approach existed anywhere in the world. Most of the problems to be solved in business are unique. Each of the issues requires in-depth analysis before we design and customize a solution for the problem. 

AI cannot solve every problem

No matter what is your problem, based on our researchers and world-class advisory team, we can analyse and discuss with you the most suitable solution for your business.  However, since we are scientists, not just someone who only want to sell you solutions, in some cases, if your problems cannot be dealt with using AI, we will also inform you rather than wasting your time and resources.

AI for Loan Approval

AI for Fraud Detection

AI for Auto Update Enterprise Architecture

  Some of Our Customized Systems... 

 Your assistance on "Making use of AI" 

While most businesses target their next move with AI as part of their engine, we, a group of AI researchers with more than 20 years of experiences, can help you identify if what you would like to do is achievable using AI technologies.

Worldwide, only around 25-40% of AI projects can really answer business problems. As a result, unlike many others, honesty is our key value, we may even tell you that "No, AI cannot help you in solving your problems".

What if AI is really your viable answer?

 

Since we are researchers, again unlike those who have their predefined solutions, you can rely on us that we will find you the best solution that AI can offer.

So, why wait?

Give us a call. Let's discuss about your wish/problems that you have in your mind, and see what can we work together.