QMulus Cloud Computing Technology

QMulus is based on Microsoft's .net platform, SQL server and the C# programming language. QMulus operates over the world wide web using standard protocols and it does not require that participating computers open IP ports for unsolicited inbound messages. QMulus uses state of the art security features to eliminate any possible adverse impacts that participants might experience.

Security Features Included In QMulus

QMulus incorporates several security features to prevent any accidental problems with your computer and to ensure that it does not create any additional vulnerability to hackers.

1. Because QMulus is working on different A.I. problems over time, Qualia Labs must deliver new code to your QMulus installation periodically. These code deliveries are both encrypted and signed with Public Key / Private Key security. This is similar to the technology used with https:\\ or "secure" web sites.

2. Data other than software updates sent back and forth between your QMulus computer and Qualia is not encrypted, to save bandwidth. However, there is no personal information here - only the results of our experiments.

3. The A.I. experiments code that Qualia loads runs in a special "protected" space on your computer (for the techies this is a .NET AppDomain). This protected area does not allow the downloaded code to have access to your disk other than a few specific directories, it cannot load other programs, etc.

4. QMulus only uses regular "port 80" internet access - it is just like your web browser. It does not open any ports or firewall areas that cause any hacker vulnerabilities.

5. QMulus is designed to automatically quit if it begins to use too much memory on your computer. Techies can adjust this in the Advanced Configuration area of QMulus.

QMulus AI Operation

QMulus enables machine learning scientists to create any number of exploratory algorithms in the name of research. The platform provides means for these algorithms to be distributed, processed, and re-assembled with very little effort. Applications in Genetic Algorithms (also known as Evolutionary Algorithms) are currently under investigation, with early efforts focused around the NEAT line of research. Many additional explorations including Quantum Evolution, Hierarchical Temporal Memories and others are expected.

ExplAIning A.I.

Ask a two year old child Where is your mommy? and you will most likely get a gratified finger extended in the correct direction. If mom is close by, you could then ask Where is mom's nose? and you would be obliged with a more precise indication, most likely with a celebratory giggle.

An A.I. researcher observing this situation faces many complex questions but they all start with a relatively simple one: How does the child separate the world into objects? And further, how does he or she know which objects form part of others? In A.I. we call this problem segmentation. Though there are many known ways to try and solve it, segmentation does not have a perfect solution.

For example, color is an easy way to separate objects. But then again, a rainbow is a multi-colored single object. Texture is another way to segment the world around us. But a bowl of apples has one texture for many individual things. Outlines or edges would help to separate each apple, but then again the corner of a box is a hard edge that does not separate two objects. Bottom line: segmentation is hard!

In fact, nature has learned to exploit the difficulty of segmentation. Zebra stripes, for example, confuse predators and stop them from being able to segment the herd into easily captured individuals. Animals of all sorts camouflage themselves to become part of other objects and thus avoid being taken out for lunch!