Branches of AI
AI is coming of age...
Machine Learning, Deep Learning, Cognitive Computing,
Robotic Process Automation (RPA), Natural Language Processing (NLP),
Machine Perception, Predictive APIs, Image Recognition,
Speech Recognition, Virtual Agent, Intelligent Assistant,
Personal Advisor, Chatbot, Semantic Search... anything missing?
If you've never heard about this term before,
just read on and you will soon understand the concept and the
benefits of the adoption of this AI technology by your organisation.
Machine Reasoning basics: Using Human 'Know-how' to Teach the Machine
Machine Reasoning (MR) systems generate conclusions from previously acquired
knowledge by applying logical techniques like deduction and induction.
Computer Scientist and Futurist Jerry Kaplan describes a reasoning system as "a concept
that deconstructs tasks requiring expertise, into two components:
1 - 'a knowledge base' - a collection of facts, rules
and relationships about a specific domain of interest, represented in a symbolic form
that a computer can understand
2 - a general-purpose 'inference engine' that describes how to manipulate
and combine these symbols."
As one of the biggest advantages of reasoning systems, Kaplan states that, as they are
based on facts and rules, the operation and performance of those kinds of systems
can be modified more easily since new facts and knowledge can be incorporated over time.
In this process, reasoning systems are taught by "knowledge engineers" who interview
expert practitioners and "incrementally incorporate their expertise into computer
programs". In this way a common language of concepts and their inter-relationships is
constructed and contained in an 'ontology'. And thus, the machine has a language it
can work with and combine to solve multiple problems on its own.
This structure and approach also makes it much more convenient to explain the
reasoning performed by the system.
What do today's Machine Reasoning Systems Look Like?
Today's machine reasoning systems feature four components:
A young child captures what it needs to learn from parents,
teachers, other children or anyone else by using its 'sensors'
like ears and eyes. For a machine, this can be done by individuals
or groups of experts teaching the machine bits of their contextual
'know-how', insight and understanding - including the what,
where, when and why (codified with assistance from knowledge
This is what the machine uses subsequently to reason as it
performs its tasks and explain its recommendations. Thus,
the AI learns best practices and reasoning from its donor
- Semantic Graph:
AI understands the domains in which it is to operate by creating
a codified semantic map of all the concepts in its world,
their inter-relationships and attributes, and the values associated
with those attributes. This 'Graph' of taught knowledge is
held within a special data store called a Triple Store. The
machine uses the contents of the store to understand the meaning
of concepts and then perform its semantic reasoning.
Example: I know that Brasil has a connection to Portugal.
And Portugal is connected to the EU. And the EU is connected
to 28 countries. And its HQ is in this buiding in a city called
Brussels where moules et frites is a popular dish. This is
a semantic graph of part of the world that we know - part
of our memory.
- Reasoning Engine:
The engine is the core service that puts everything together
and delivers a solution to a certain problem. It accesses
everything it needs know from its Triple Store and finds the
correct solution to a specific problem on its own, step by
step, based on the knowledge it has access to.
- Adaptive Problem Solving:
Adaptive Problem Solving, also known as Machine Reasoning
(MR), is the ability to dynamically react to change and by
doing this, re-using existing knowledge for new and unknown
problems. With MR, problems are solved in ambiguous and changing
environments. The AI dynamically reacts to an ever-changing
context, selecting the best course of action. Thus, MR forms
the basis for a general artificial intelligence that can adjust
its recommendations, according to circumstance, just as a
human expert would.
Integrating Machine Reasoning and Machine Learning
As summarised by Kaplan, "...if you have to stare at a problem and think about it,
a Machine Reasoning approach is probably more appropriate. If you look at lots of
examples or play around with the issues to get a "feel" for it, Machine Learning is
likely to be more effective."
The good news is that real-world AI solutions require BOTH these forms of AI working
In this enhanced form of AI, MR performs the logical reasoning and leads you through
the steps of a complex process as it adapts to real time change, while specialist ML
routines provide the best contextual data values to be used by the MR functions as they execute.
Recommended reading: Jerry Kaplan's
"Artificial Intelligence: What everyone needs to know"
is probably the best book right now for those who'd like to read further.