[attachment=12271]
HIERARCHICAL TEMPORAL MEMORY
 WHY CANT COMPUTERS BE MORE LIKE THE BRAIN?
WHAT ARE HTMs?
 A machine learning model that replicates some of the structural and algorithmic properties of the human brain.
WHY NEED HTMs?
 No viable algorithm to perform cognitive tasks such as visual pattern recognition and understanding spoken language on the computer.
 In a human, these capabilities are largely performed by the neocortex – a thin sheet of cells, folded to form the convolutions occupying nearly 60 percent of brain volume.
STRUCTURE OF HTMs
HOW DO HTMs WORK?
 Works on a common algorithm for all modalities of sensory input.
 Knowledge distributed across the nodes up and down the hierarchy.
 Simple structures in low-level nodes and more complex structures in higher-level nodes.
CAPABILITIES OF HTMs
Capable of performing 4 basic functions :
1. Discover causes in the world
2. Infer causes of novel input
3. Make predictions
4. Direct behavior
WHY IS HIERARCHY IMPORTANT?
 Shared representations reduce memory requirements and training time.
 Mirrors the hierarchical structure of the world (both in space and time)
 Belief propagation-like techniques ensure the network quickly reaches the best mutually consistent set of beliefs
 Affords a simple mechanism for covert attention
 Belief Propagation
 Each node in the system represents a belief that is mutually consistent with all the other nodes.
 Resolves ambiguity.
 Makes large system settle rapidly.
IMPLEMETATION DETAILS
 Current version of software platform is the research release called ‘Numenta Platform for Intelligent Computing (NuPIC).
 Available totally free of cost.
 Composed of three main components:
1. Run Time-Engine.
2. Development Tools.
3. Plug-in API and its associated source code.
 A set of sample HTM networks, as well as documentation and training materials also created to help the developers get started.
 Written in the Python scripting language.
 Source code available for inspection for developers to modify and enhance them.
APPLICATIONS OF HTMs
 Car Manufacturing – visual recognition of various parts.
 Modeling Networks, including computer networks, power networks etc – predicts undesirable future conditions.
 Oil exploration – to decide where best to drill a well.
 Pharmaceutical firms – discover and test new drugs.
 Businesses – study corporate behavior.
 Robotics – direct the actions of an android.
FUTURE ENHANCEMENTS
Being a new technology, there are many advances ahead in our understanding of HTMs
1. Improve ability to measure and define the capacity of an HTM.
2. Develop useful heuristics for how best to specify hierarchies to match particular problems.
3. Improve the existing algorithms for spatial quantization and time-based pooling.
4. Extend the implementation of Numenta HTM platform to Windows-based systems.
CONCLUSION
 We have recognized a fundamental concept of how the neocortex uses hierarchy and time to create a model of the world and to perceive novel patterns as part of that model.
 If we continue the advancements and refinements in the right direction, the true age of intelligent machines may just be getting started!

Prepared by
Jasna.A

---

[attachment=14934]
HEIRARCHICAL TEMPORAL MEMORY
HTM
A machine learning model.
Developed by Jeff Hawkins & Dileep George.
Is a Biomimetic model .
Based on memory-prediction theory of brain function.
What is HTM??
Heirarchical temporal memory.
That models the neocortex.
Is a memory system that learns about its world by being exposed to sensory data over time and abstracting high-level concepts from data.
HTM network can predict what is likely to happen next.