Deep learning

An Inquiry voyage
Dr. Joseph Shapira
jshapira@netvision.net.il

What shall we talk about?

About learning laws
and understanding systems
What is scientific intuition
Why motivation
What is inquiry
and creative engineering
2

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

What is understanding
Understanding is building a story
that links a new experience and inference to existing cognitive
picture of the world - and expands it.
Understanding is introducing a new animal to
Alices Wonderland
Understanding is actively solving new problems.
Passive understanding
is only the satisfaction of listening to teacher/reading textbook.
The trust in the book/ teacher waives the need to question/ inquire.
This is rote learning.
3

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

And what does it take to understand?

A world of experienced memories and concepts encompassing the
personal secured space.
Linkage of the new experience to memories having some similarity
Note: 95% of the memories are not verbal or written equations but audio,
visual, haptics, smell, etc. The multimedia sensory memories provide
multiple links to the new experience.

Imagination visioning the scenario within the familiar world,

examining and comparing.
Critical thinking
Independent tests by comparison to known scenarios
Logic reasoning

Imprinting in the cognition ( enhanced by motivation)

4

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Intuition
Understanding is conceiving a model that predicts the phenomena. In
search for the rules affecting the scene, and their balance, the brain
prefers to guess and estimate similarity over running an infinite
sporadic search.
This guess, relying on previous similar experiences, is intuition
- The unstructured thinking invoking relevant associations.

Every human creation involves intuition,

thus differing from machine thinking.
5

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

When intuition meets a conflict

A child is born with the guess instinct, but his intuition is built through
accumulation of experiences.
When the intuition conflicts with the observed situation:
It is recalibrated if the conflict is surmountable. A richer intuition is
developed.
It is destroyed ( looses credibility) if the conflict is essential.

Intuition is essential for understanding. How do we nurture

intuition:
By creating consecutive surmountable conflicts that tame the
intuition to incorporate the new experiences and inferences.

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Motivation
The brain, as a sentinel, observes situations, predicts threat/ opportunity and
commands action.

Motivation is the prize for performing a challenging command.

It is an intrinsic function of a living system ( remember Pavlov?).
Why is motivation essential for learning?
Motivation energizes the body and brain systems and imprints the
experience and the lesson learned.
The challenge in science research: exploration, conception and then testing
models is rewarded:
by expansion of the comprehension of the world ( the enlightening)
By being appreciated by close friends.

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Adapting to a Changing World

The Committee on Undergraduate Physics Education Research and
Implementation, National Academies

An overarching theme has emerged from educational

research: Learning improves when students are
interactively engaged with their peers, their instructors,
and the material being learned, and when they are
integrating the newly learned concepts with their previous
ideas, whether learned in a formal classroom or in
everyday life.
http://www.nap.edu/catalog.php?record_id=18312

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Einstein
and the importance of the mental picture
From Einsteins Biography ( Walter Isakson 2007):
During his 16s Einstein attended highschool based on Pasteluccis
philosophy that encourages the pupils to imagine pictures,
extract their own inferences from observations, develop
intuitions, conceptual thinking and visual imaging

The visual understanding of concepts became a

substantial aspect of his genius
this kind of visual Gedankenexperiments became a
symbol of his career

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Physics teaching: passive vs active

Objectives:- Productive and creative Skills and values
- Expertise in the field of knowledge
Passive learning (teacher-board-class)
Formal (mathematical)and general definition of laws of nature
Exercise: Data (given), laws (known), results ( to be solved)
Controlled imaginary scene incorporating few laws.
The learning process serves operational skills, not creativity
Non-real scenery and parameters impedes estimation, analogy
and critic, and does not contribute to scientific intuition
10

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Richard Feynmann
Nobelist physicist and an esteemed teacher
I can't understand anything in general unless I'm carrying
along in my mind a specific example and watching it go
Understanding from the actual case to abstraction

"Physics is not complete

We do not know yet how the machine works
There is no axioms that reality has to match to,
But rules derived from observations and analysis

11

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Inquiry-based/creative learning
Inquiring familiar environment
Observation (assembled), Controlling rules ( not known, needs
creativity), Behavior ( observed)
Conceptual inference of laws of nature
Critic through analogy, estimation, validity
Formal mathematical formulation
The process proceeds from personal experience, related to past experience
, through imaginative creation, estimation and critique (intuition).
Observation of reality involves multiple details and multiple senses, that
lead to previous relevant experience and inference.
This is how scientific intuition is built.
12

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Inquiry of systems
Cause and effect

Logical/ causal axis

Structural axis

Component and
structure

Top-down

Forward Rules

Backward

Structure

Bottom-up

13

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Analysis of systems
Complex systems are not transparent
System inquiry is seeking the topology that links cause to
the observed effect.
Logic/ causal axis (balance and sequence of rules)
Observed effect
Possible source/ rule
Structural/ geometrical axis
Observed structure
Possible building blocks
The course of inquiry is a travel back and forth in both axes:
From observation to laws, from the whole to components
14

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Physical laws and systems

Laws are the building-blocks of the systems of nature,
understanding of which is the ultimate goal
Learning and assimilating laws Exploring/developing systems
The system is defined by the inquiry
question, or by its operational objectives.
From authentic case to generalization Observation and analysis according to
the impact on the systems performance.
From living reality to abstraction
Intuitive guess is intrinsic in the process.
Assimilation through the detailed
Predictive estimates are essential for
convergence.
inquiry process
From experience to the concept

Dr Joseph Shapira
Deep
15
Learning
1 Aug. 2015

The inquiry tool-box

Conservation laws
Scaling (validity of)
Parametric response ( structure, time-line), and limitsd on
validity of the model.
Dimensional analysis
Sequential , hierarchical modeling, each approximated to
the accuracy required by mother model.
Analogy
Imagination
16

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

On hypotheses and estimates

Things should be made as simple as possible ,

but not any simpler.
Albert Einstein

"Plurality should not be posited without necessity

William of Occam

(Between equally plausible hypotheses, the one with minimum posits should survive)

When things get too complicated, it sometimes makes sense to stop and wonder:
Have I asked the right question?
Enrico Bonbieri

The buildup of a model is an iterative sequence of guess, estimate a

model and test, correct.
The guess is intuitive, based on relevant accrued memories.
17

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Physics is the platform

Physics is the mother of sciences.
The childs neighborhood familiarizes him with physical phenomena.
A small number of rules enables analysis, modeling and prediction.
Learning through authentic examples, and personal experience
builds intuition, assessment, critic, and linkage between events.
Technology is an integration of physics laws with a strategic goal.

Physics learning is the platform to conceptual

understanding of systems in nature and technology

18

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Training of inquiry-teachers
The paradigm change from structured passive teaching to cooperative
inquiry is a conceptual transformation, requiring mental readiness.
The creation dominates. The process evolves through daring, exposure,
cooperation and peers reflection.
Exposure, integrity and trust are the key words.
The value of the inquiry is important to the teacher as it is to the student.
The teacher is empowered and creates, and enlightens the students.
The teacher needs a supporting environment,
similar to the engineer and the resercher.
A reflecting and debating teachers community is needed.
The involvement of experienced engineers, is essential, for bridging the
cultural gap.
19

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Teachers communities
The concept of teachers communities is spoken in the pedagogical
community,
Its implementation does not have a marvelous record.
Why teachers communities?
The role of a teacher as a curiosity, creativity and critical thinking exciter
and empowerer, and knowledge organizer, and as a bridge to the world
of science and technology is creative and demanding more than that
of an average engineer or scientist.
Yet, the teacher does not have a structured frame for communicating,
updating, debating, peer reflection as the engineer and scientist have.
A science-oriented peer community is needed to revive and maintain the
teachers striving for excellence.
20

Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Achievement goals - summary

Assimilation of deep learning
Development of critical thinking, analyzing, comparing, estimating
and creating.
Development of scientific intuition through continuous linking new
experience and inference to the mental picture of the world.

Broadening the tool-box of concepts and tools

Physical, mathematical, engineering

Creative development of a system

21

Definition of purpose and its operating environment

Acquaintance with the relevant laws, tools and integration processes
Project planning for effective risk reduction
Experiencing creativity
Critical tests
Dr Joseph Shapira

Deep Learning

1 Aug. 2015

Thank you
for your attention
D r J o s e p h S h a p ira
0546 607088

js h a p ira @n e tv is io n .n e t.il

22

Dr Joseph Shapira

Deep Learning

1 Aug. 2015