Frequently Asked Question
AiSara . Machine Learning . Technologies
What is AiSara?
AiSara is an artificial intelligence technology for analytics, regression, prediction. So far we have made it available as an Excel Add-in. Futre plan includes making its API available for software/services on demand and extending its capability. Can try our demo to play around and understand more on how AiSara works
What does AiSara stands for?
'Ai' stands for 'Artificial Intelligence'.
'Sara' is an acronym for 'Solution Approximation with Robust Algorithm'.
What you can do with AiSara?
- Understand your complex data.
- Understand cause and effect.
- Understand relationship between different variables.
- See where is the maximum and minimum of your data.
- Understand the relationship between your inout data and output data you are observing.
- See all possible solutions.
- Understand if there are alternative solutions.
To explore AiSara capability, download data from Sample Dataset.
How is AiSara different than other technology?
AiSara is so easy to use that we can provide it as an Excel Add-in. If you know how to use Excel, you will know how to use AiSara quickly.
Can you give examples of use?
Sure.
Predictive Maintenance: Input data from your equipment, machinery, facility etc. and obtain data from sensors attached to them. Why? This allows the detection of wear and tear, scheduling just-in-time (JIT) maintenance to avoid costly repairs and expensive downtime.
Business Sales: Input data being customer, products, branch, services, marketing effort or in other words, who, what, when, where and how much. Output data being sales, revenue, profit. Why? Analyse and learn customer behavior, maximizing sales and serving your customers better. Anticipate the increase in demand or a changing trend. Know what you don't know yet.
Oil and Gas Subsurface Studies: Understand complex data from subsurface studies. Input data being rock properties such as permeability, fault transmissibility and other factors such as reservoir size. Output being the error or objective function of your study. See what influences the results the most and where the solution may lie, or where you need to look for it.
Medical: Understand patients' health data, and diagnose and predict accordingly.
Strategic analysis: Large companies or countries can run economic, political, and various scenarios.
Scientific and Research Development: Test impact of different parameters or combination of parameters on output.
See our Sample Dataset
How can I use AiSara?
The simplest way is to drag and drop your data within AiCentrica into AiSara.
In the future we will make APIs (Application Programme Interface) available.
How does it work?
If you haven't seen the video in our main page, we encourage you to do that first.
What we have done is taught computer how to see a pattern like a human does. It can see repeating patterns and know what is noise and what isn't. It can see what is important to the output and what is not. Humans can do much more of course such as contextual information, but what we have done is allow the computer to see how the parameters affect the results. Humans can only see 3 dimensions, but once we have taught the computer to see the pattern, it can do it in 3D, 5D, 10D or more. Lots of data application is complex, this allows for quick understanding which we may miss and it does the analysis in a short time frame.
There is much comfort in being able to see how AiSara performs, you can compare the data visually with how Aisara sees the pattern near and away for the data.
When we can relate the input to the output quantitatively, we do not need to use equations, and for many use cases, there are no equations. That is why suddnely there is so many use cases.
The orange points below are data provided. The image below shows how AiSara is able to fill in the missing pattern.
Can I write my own Excel macros utilizing the LEARN and PREDICT functions?
Yes, of course!
How did you guys figure this out?
We have been looking at some of the most complex and uncertain data in the world. The creation is driven by our needs to solve the problem in oil and gas but we find that the technology can be applied to many sector/industry. AiSara looks for pattern in the data and there is pattern everywhere.
How AiSara different from Deep Learning?
Essentially, deep learning uses linear algebra arrays/matrices/tensors to numerically relate input data to the output. There is more to deep learning than this but we try to keep it simple here. You can pull apart the tensor and scratch your head to understand what it means, (which parameters are important and how it affects the output) but in many cases it is not practical or actually impossible.
With AiSara you can explore the solution space and with 3D feature in the future, literally see the whole solution space. You can compare the actual data to the solution provided by AiSara and judge the accuracy of AiSara for yourself.
My company has so much product data and combination of products for different usage. Can I use this to show my client the best product for their specific need?
Yes. Of course. This can also work if you manage a database and your client needs to find the output (product?) given certain combination of input.
Can I use this for my research on fluid dynamics or weather prediction?
We welcome you to try!
Where are you guys located and how can we discuss possible partnership, collaboration, investment?
We are located in the heart of Kuala Lumpur, Malaysia. We welcome further enquiry at support@aisara.ai
My Company has an app/wearable but no analytics. Can we collaborate?
Definitely. By complementing your app with AiSara, you are stretching the limitations of your app, unlocking endless possibilities. Get in touch with us at support@aisara.ai
is AiSara for?
AiSara is for analysts, engineers, researchers, consultants, students, educators and all professionals from different disciplines.
Are you using neural network or GAN?
No, we are using our proprietary pattern recognition technology. The reason being we need a technology that allow us to display all the solution space, local minimas, local maximas. We need a technology that is interperetable, explainable. Say for example, I can visualize where the data is and how good the solution is compared to the data, and how the different variables impact the result.