Introducing the GibiNeuron!

The following are terms to describe the number of neurons in a neural network.

Metric prefix

Unit	Shorthand	Power	Number of Neurons
Neuron	N	1 Neuron
Kiloneuron	KN	1 0001	1,000 Neurons
Meganeuron	MN	1 0002	1,000,000 Neurons
Giganeuron	GN	1 0003	1,000,000,000 Neurons
Teraneuron	TN	1 0004	1,000,000,000,000 Neurons
Petaneuron	PN	1 0005	1,000,000,000,000,000 Neurons
Exaneuron	EN	1 0006	1,000,000,000,000,000,000 Neurons
Zettaneuron	ZN	1 0007	1,000,000,000,000,000,000,000 Neurons
Yottaneuron	YN	1 0008	1,000,000,000,000,000,000,000,000 Neurons

Binary prefix

Unit	Shorthand	Power	Number of Neurons
Neuron	N	1 Neuron
Kibineuron	KiN	1,0241	1,024 Neurons
Mebineuron	MiN	1,0242	1,048,576 Neurons
Gibineuron	GiN	1,0243	1,073,741,824 Neurons
Tebineuron	TiN	1,0244	1,099,511,627,776 Neurons
Pebineuron	PiN	1,0245	1,125,899,906,842,624 Neurons
Exbineuron	EiN	1,0246	1,152,921,504,606,846,976 Neurons
Zebineuron	ZiN	1,0247	1,180,591,620,717,411,303,424 Neurons
Yobineuron	YiN	1,0248	1,208,925,819,614,629,174,706,176 Neurons

Amazing program "YOLO"

Y.O.L.O., acronym for "You Only Look Once" is a new machine learning architecture developed by Joseph Redmon that allows a program to detect and classify objects from video streams in real time.

Let him present it further himself:

Short introduction to HTM

Jeff Hawkins is an engineer, business man, neuroscientist and inventor who spent his life in the pursuit of a grand unified theory about how our brain works. He has a noteworthy list of accomplishments behind him, and is very successful in his quest.

I first heard about him when I watched his ted talk, and from there my curiosity took over, and I started looking more into the research he is conducting with his company numenta. I will try to summarize the basics of his theories in this post.

If you want to learn more, a good place to start is to read his book "On intelligence".

So what is HTM? HTM is an acronym for Hierarchical Temporal Memory and basically it is a list of 3 important aspects of the algorithm in the theory. The most prominent contributor to intelligence in our brain is the "neocortex".

In contrast to the rest of the brain which has evolved longer and therefore is much more specialized, the relatively new neocortex has a homogenous structure that is re-used throughout.

This structure is basically that the cortex is a 2mm sheet of neurons about the size of a large napkin, crumpled to fit inside our cranium. The sheet is divided into 6 layers of neurons that are connected in a particular way.

Small patches of the cortex represent stages in a hierarchy, and the connections between the neurons is what dictate the boundaries of each patch. Signals pass from lower stages up to higher stages and back down.

The lowest stages are connected to our senses such as eyes, ears and skin via mechanisms of the old brain. The signals passed from here consists of patterns of impulses that make their way up the stages, and it is temporal sequences of these patterns that are learned in each stage.

And perhaps the most important part of the theory is the following; Each stage will try to remember signals coming from lower stages and later predict those signals once they have been remembered. Only the signals that have not been predicted or "learned" will pass to the next stage unchanged, so by every stage the impulses have been refined and condensed into more abstract and high-level information.

 When a stage is seeing a new signal that it cannot predict, it will be sent up to the next stage until one stage recognizes it. If no stage recognize the input, it is forwarded to the hippo-campus which sits at the logical top of the hierarchy. The hippo-campus will keep the unknown information around for a some time until it is no longer useful. Hopefully, the stages below will now have managed to learn it, and if they have not, it will simply be discarded.

Beyond this introductory description of HTM, there are many important details that really describe well how this relatively simple algorithm actually explains our intelligence and sense of being completely.

I can warmly recommend to read the book "On intelligence".

tiny-cnn

I know that Deep Learning is the future of anything related to intelligent robotics. Since 2012 it is the most disruptive technology in the field, making several decades worth of carefully hand engineered and tweaked code bases obsolete literally over night. And as a result, all the big players in IT such as Google, Facebook, Nvidia etc. are pouring their biggest bucks into this retro area of research.

Retro? Yes, because research has been done for decades in this field before it was kind of forgotten. Why was it forgotten? After numerous stabs at getting a working implementation of the many advance models based on biological principles like neurons and synapses, it was collectively deemed too resource intensive for the contemporary computer hardware.

But along with fancy virtual reality headgear the "neural network artificial intelligence" of 1980's and 1990's sci-fi has now resurfaced, this time with actual real promise (For VR see Oculus Rift).

How? Well because of the unfathomable increase in computer's capacity to process, store and communicate data combined with the unfathomable increase in the number of computers connected together in easily accessible clusters and farms such as AWS combined with maybe the most important parameter: the unfathomable amount of readily tagged media for training purposes (read: cat pictures on youtube).

NOTE: These graphs really do not not give my statement justice unless you grasp that the scale is exponential, and notice the part where it says "human brain" to the right.

Suddenly the dusty old models from the 1990's could be plugged into a new computer and give results literally over night.

I am truly fascinated by this new-old technology that suddenly promises that our computers in the near future may understand our desires in an exponentially growing degree. A technology that makes self-driving cars intelligent speech driven assistants something that we get to see not sometime  before we die, but something we can buy within a decade or even sooner. And who knows what kind of crazy tech we will depend on once this first generation of deep learning has reshaped our lives?

I am a true beginner in Machine Learning and Deep Learning, and I intend to use OctoMY™ as a vessel for learning about it. It is my ambition to make DL an integral part of OctoMY™ as soon as possible, putting it in the hands of the hobby robotics enthusiasts out there. Because, as fascinating as DL is, almost no-one outside the field understand what it is, the significance it carries.

But where would a beginner like myself start? It is really a jungle out there, in terms of available software libraries/frameworks/toolboxes that promise a varying degree of features, performance and integration.

So, in my quest to find the perfect deep learning library/framework/toolbox to use from OctoMY™ I found this useful presentation of tiny-cnn, and I decided to try it out.

According to the project page on github tiny-cnn is

A header only, dependency-free deep learning framework in C++11

 It promises integration with data developed in Caffe. I will give it a shot and we will go from there. Stay tuned!

Software stack schematic

As in my previous post, here is a draft of how I plan to lay out the software in the DEVOL robot.

DEVOL Software Stack schematic draft

It can be broken down to the following components:

Audio and Video inputs are filtered through a graph of detectors, recognizers and other more or less involved processes that translate them into useful events such as people, objects, facial expressions, distances, location and so forth.

These events are gathered together with sensor inputs in an event inventory where all events are classified, sorted, persisted, enriched and refined in real-time.

The core of the system is the "consciousness driver" which is simply the main loop of the program. It relies heavily on its array of close assistants that are responsible for their respective expertises of keeping up communications with HQ, inducing logic , keeping track of objectives, keeping track of ongoing dialogues, keeping track of appearances in form of pose and avatar and so on.

The consciousness driver will base its decisions on the content of the event inventory and its decisions will result in changes to pose, additions to the local logic facts database, output of audio dialogue and display of avatar on screen.