Context

#

Learned in this study

#

Things to explore

#

Overview

#

Why will machines intelligence be based on cortical principles?

#

• The cortex uses a common learning algorithm
• The cortical algorithm is incredibly adaptable
• It has a network effects: hardware and software efforts will focus on the most universal/generic solution

Cortical facts

#

Sensory systems:

• Retina
• Cochlear
• Somatic

Patterns of action potential, firing of neural fibers. The brain deals with patterns. The neocortex learns a model from fast changing sensory data. With the model, it can generate

• predictions
• anomalies
• actions

The neocortex learns what is called a sensory-motor model of the world. The brain is a sheet of cells which is remarkably uniform. It is organized as a hierarchy. Within a level of the hierarchy are cellular layers. Within those layers there’s an organisation called mini-columns. At the end are neurons. Learning is about modifying synaptic weights, but also about degenesis/neogenesis of synaptic connections.

Cortical theory (HTM)

#

Overview

#

1. Hierarchy of identical regions
2. Each region learns sequences (time-based patterns)
3. Stability increases going up the hierarchy if input is predictable
4. Sequences unfold going down

Questions

#

• What does a region do?
• What do the cellular layers do?
• How do neurons implement this?
• How does this work in hierarchy?

Cellular layers

#

1-6 layers, 2 to 3 layers of feed forward (2-3-4) and 2 layers of feedback (5-6) Each layer is implementing a type (variation) of a common sequence memory algorithm Layers 2-3-4 are doing inference Layer 5 is doing motor behavior Layer 6 is doing attention

The input to a particular region arrives at the layer 4 (L4) and then projects to L3 which then projects down to the lower layer of the hierarchy The motor behavior is also passed in at the same time Thus, what is received is the information that is perceived as well as the recent behaviors of the body

L4: Learns sensory-motor sequences If the layer is able to predict properly the sequence, it forms a stable representation that is passed onto L3 if it is unable to predict, it passes through the change to L3

L3: Learns high-order sequences

The neuron

#

Biological neuron

#

10% of the synapses are close to the cell body Feedfoward input Added linearly Generate spikes

2 regions Basal dendrites (bottom, close to the cell) Apical dendrites (top, far from the cell) They are non-linear Dendritic action potentials depolarize soma

HTM neuron

#

Feedforward Linear summation Binary activation

Distal synapses Modeled as a set of coincidence detector Threshold coincidence detectors Puts the cell in a predicted state

Biological synapses

#

Learning is mostly formation of new synapses Synapses are low fidelity

HTM synapses

#

Scalar permanence Binary weight

Sparse distributed representations (SDRs)

#

Called The language of intelligence

Many bits (thousands) Few 1’s, mostly 0’s Each bit has semantic meaning Learned

SDR properties

#

1. Similarity shared bits = semantic similarity
2. Store and compare store indices of active bits subsampling is OK
3. Union membership can ask “is this pattern part of the union?”

A cell can recognize many unique patterns on a single dendritic branch

A cell activates from dozens of feedforward patterns It predicts its activity in hundreds of contexts

Learning transitions

#

1. Feedforward activation
2. Inhibition which generates sparse cell activation
3. Formation of connections with nearby cells which were previously active cells (give them the ability to predict future activity)

If a pattern is input, many cells will indicate that they predict to be activated next It can predict A-B, A-C, A-D This is known as a first order sequence memory It cannot learn A-B-C-D vs X-B-C-Y Mini-columns turn this into a high-order sequence memory

Forming high-order representations

#

If there is no prediction, all cells within a column become active If there is a prediction, only the predicted cells will become active (the other will be inhibited)

HTM temporal memory (aka cellular layer)

#

Converts input to sparse activation of columns Recognizes and recalls high-order sequences

Desirable traits

#

• Continuous learning
• High capacity
• Local learning rules
• Fault tolerant
• No sensitive parameters
• Semantic generalization

Research roadmap

#

Applications using HTM high-order inference

#

graph LR;
	0[Data]
	0 --> 1[Encoder]
	1 --SDR--> 2[HTM high-order sequence memory]
	2 --> 3[Predictions<br/>Anomalies]

Thoughts on machine intelligence

#

• Cortical (HTM)
• ANNs (deep learning)
• AI (Watson)

See also

#

References

#

• Video: https://www.youtube.com/watch?v=6ufPpZDmPKA
• Slides: http://www.slideshare.net/numenta/2014-10-17-numenta-workshop