Tensorscience.com is an open-source and free repository of guides and tutorials that use deep learning to solve complex challenges, such as those involving object recognition, optical character recognition and natural language processing. It aims to link the science of machine learning on the one hand - which in certain areas is still theoretical and out of the realms of current computer capabilities - and software and chunks of code on machine learning on the one hand that are available on the internet. The primary programming language on Tensorscience.com to conduct machine learning is Python.
If you have questions, suggestions, or other inquiries, please contact us through our Contact page.
The knowledge on machine learning and its capabilities has existed for decades but has only become graspable in the past decade with innovations in computer hardware (e.g. gradient processing capabilities in GPUs) and software. This is a very exciting development. Humans have for the most part just crawled out of the slime. In the last four hundred years, human cognition has delved into realms previously inaccessible by any other creature in the last billion; complex and abstract mathematics, chemistry, physics, and recently, computer intelligence has transformed our world.
Our recent ancestors didn’t know how a television worked, and only rode on an airplane a few times in their lifes. Today, children are walking up to televisions and swiping right in frustration. Our very brains are changing due to the heightened pace afforded by technological advances.
In the matter of a few decades, humans will cease to exist in the way we have for the past 4 million years. The advent of server side architectures transacting everything from commerce to entertainment shapes our lives, changes our environment, and recommends what we should do next. Human machine interfaces are becoming ubiquitous and will, hopefully, be important drivers of human progress in the years to come.
The thin veil that makes all this possible is the internet, and the machine intelligence that shapes it. So far, this intelligence is embryonic; business rules are not bonafide intelligence, and as such, most applications today are governed by static inputs and outputs. A google result is very useful for retrieving information, yet it fares very poorly in actually answering your question.
So what then is intelligence?
Using a broad brush, the following might be considered an al fresco mural defining intelligence,
- the capacity to acquire and extract knowledge from observations,
- the ability to respond and apply knowledge given observations,
- the capacity to understand and foresee the intention of other systems
- the ability to recognize and draw (plausible) inferences from observations
- and most importantly, posing questions and trying to answer them
Machines are getting closer and closer to being able to do this. We hope that the guides on this website will help you in learning from, and applying, such artificial intelligence.