Cognitive Sciences Related Knowledge

Cognitive Sciences Related Knowledge

Developmental & Learning Psychology

Linguistics 🗣 🔤 🈳

Areas of research

Neurolinguistics

Language Development

Psychology 💬 🧠 🗨

Views/Approaches

U-Shape Pattern of Development

Language Acquisition

Begin with copying/imitating

Functionalism

Historical linguistics

Grammatical Rules

Two-Word Stage

Characteristics Of Language

Structuralism (1880-1900/60)

Voluntarism

Psychoanalytic Psychology

Communicative

Dynamic

Babbling Stages

Morphology

Phonology

Structured

Learning exceptions of rules

Understand rules of grammar

One-Word Stage

Syntax

Arbitrary

Philosophy 📚 💭 🤔

Philosophy of Mind

Anthropology 📿 👥 📝

Behavioural Economics

Loss Aversion and Framing Affect

Cognitive Anthropology

Neuroscience 🐀 🔬 🧠

Artificial Intelligence 📊 🤖 📉

Evolutionary Psychology

The Mind Body Problem

Neurobiology

Mental Accounting

Knowledge-Acquisition Problem

Evolution and Judgment under Uncertainty

Dualism

Evolution and Natural Selection

Neuroanatomy

Consciousness

Neurons

Nativism

Comparative Cognition

Loss Aversion

Fallacies

Associationism

Structures

Types of Knowledge

Rationalism

Property Dualism

Monism

Evolution of Logic and Reasoning

Myelin Sheath

Evolution and AI

Approaches

Substance Dualism

Declarative

Empiricism

Philosophy of Science

Concepts

Cortical Features

Metaphysics

Study of Human Origin through evolution

Sunk-Cost Fallacy

The Chinese Room Argument

Soma

Gambler's Fallacy

The Watson Selection task

Heuristics

Animal Problem Solving

Functionalism

Transitive Interference

Cell Body

Directions in the Nervous System

Nucleus

The Ultimatum Game

Idealism

Scientific method

Evolution and Language

Base Rate Fallacy

Synapse

Endowment Effect

Methods - Logic and Reasoning

Classical Dualism

Posterior "Back"

Procedural

Variation

Cerebral Hemispheres

Medial "Middle"

Neural Proces

Dendrites

Axon Terminals

Epistomology

Cartesian Theatre

Evolution and Sex Differences

Ventral "Bottom"

Comparative Neuroscience

Object Permanence

Selection

Hard Version

Dorsal "top"

Inheritance

Easy Version

The Multiple Draft Model of Consciousness

Human brains compared to other animal brains in terms of size, structure, and cognition

Physicalism

Cephalization Index

Human Computer Interaction 👤 ↔ 💻

Education 🎓 💡 👶

Psychology

Artificial Intelligence

Neurosciences

Human Computer Interactions

à organiser

Anthropology

Linguistics

Philosophy

Education

Engineering

Type of Interactions

Haptics

Visuals

Machine Learning Models

Machine Learning - Approaches

Deep-learning Algorithms
Concerned with building larger and complex neural networks, with many of them concerned with semi-supervised learning problems. These are updated form of Neural Networks

Deep Boltzmann Machine - DBM

Convolutional Neural Network - CNN

Deep Belief Networks - DBN

Bayesian Algorithm
Methods that apply Bayesian Theorem for problems such as classification and regression

Bayesian Belief Network (BBN)

Gaussian Naive Bayes

Naive Bayes

Bayesian Network (BN)

Multinomial Naive Bayes

Decision Trees
Decision tree methods construct a model of decisions made based on actual values of attributes in the data. Decision trees are trained on data for classification and regression problems.

Iterative Dichotomiser 3 (ID3)

Classification and Regression Tree (CART)

Decision Stump

Conditional Decision Trees

Chi-squared Automatic Interaction Detection (CHAID)

Ensembles
Methods that learn from weaker models

Gradient Boosting Machines - GBM

Boosting

Bagging - Bootstrapped Aggregation

AdaBoost

Random Forest

Gradient Boosting Regression Trees - GBRT

Instance-based Algorithms
Such methods typically build up a database of example data and compare new data to the database using a similarity measure in order to find the best match and make a prediction. For this reason, instance-based methods are also called winner-take-all methods and memory-based learning.

kNN - k Nearest Neighbours

LVQ - Learning Vector Quantization

SQM - Self Organizing Map

LWL - Locally Weighted Learning

Artificial Neural Network
These are inspired by the structure and function of human brains. Such algorithms, determine the importance / weights for each variable to determine the output for a given incident

Types / Cost Function

Radial Basis Function Network (RBFN)

Back Propagation

Hopfield Network

Perceptron

Regression Algorithms
In such algorithms, the relationship of the variables (Dependent and Independent) is iteratively refined using a measure of error

Ordinary Least Squares Regression

Multivariate Adaptive Regression Splines (MARS)

Linear Regression

Logistic Regression

Stepwise Regression

Locally Estimated Scatter plot Smoothening (LOESS)

Regularization Algorithms
These are similar to regression methods that penalizes models based on their complexity, favoring simpler models that are also better at generalizing

Least Angle Regression (LARS)

Ridge Regresion

Least Absolute Shrinkage and Selection Operator (LASSO)

Elastic Net

Association Rule Learning Algorithms
Such algorithms generate rules based on historical records, for eg: Whenever someone buys bread in a store, quite often people buy milk along with it. Such association rule are generated using such algorithms

Eclat algorithm

Apriori algorithm

Clustering Algorithms
Grouping values together based on their distance with each other in a multi dimension space

k-Medians

Partition Around Medioids (PAM)

K-Means

Hierarchical Clustering

Stacked Auto-Encoders

Maths

NN & DL papers

Neural Net Arch Genealogy

CNN

Semantic Segmentation

Super-resolution

Object Detection

TTS

SENet: Squeeze-and-Excitation Networks, '17.09

ResNet, '15.12

GoogLeNet, '14.09

VggNet, '14.09

Generative Models

Latent variable models

Variants

Applications

Autoregressive models

PixelRNN, '16.01.25

PixelCNN++, '17.01.19

MADE, '15.02.12

NADE, '16.05.07

PixelCNN, '16.06.16

Memory Networks

Neural Programming

Neural Programmer,'16.08

Neural Random-Access Machines,'16.02

Hybrid Computing, '16.10

Neural GPUs Learn Algorithms, '16.03

Hierarchical Attentive Memory, '16.02

Neural Module Networks, '16.06

Neural Turing Machine,'14.10

End-to-End Memory Network,'15.03

Memory Networks,'14.10

DMN: Dynamic Memory Network, '16.03, DMN+, '16.04

Reinforcement Learning Algorithms

RNN

S2S: RNN Encoder-Decoder, '14.06

Attention: Jointly Learning to Align, '14.09

Effective Approaches to Attention, Luong et al. '15.08

Transformer: Attention Is All You Need, '17.06

DCN: Dynamic Coattention Networks, '16.08, DCN+, '17.08

ACT: Adaptive Computation Time, '17.05

GRU, 14.11

DeconvNet

Data

Informatics

Ethics

Databases

NN & DL papers

Applications

Bioinformatics

Linguistics (NLP)

Robotics

Computer Vision

Math for Machine Learning

Matrices 📑

Matrix product properties

Matrix Products

Distributativity
A(B+C) = AB +AC

Not commutativity

AB!=BA

Associativity
A(BC)=(AB)C

The Identity Matrix
IA=A

All ones on the diagonal

Properties of the Hadamard Product

Commutativity
AoB=BoC

Associativity
A(BC)=(AB)C

Distributativity
A(B+C) = AB +AC

Linear dependence

Definition
lies in lower dimentional space
if there are some a-s, that
a1*v1+a2*v2+...+ak*vk=0

Example

a1=1, a2=-2, a3=-1

det(A)=0 only if columns of A are linearly dependent

Dot product and how to extract angles

Hyperplane - is the thing orthogonal to a given vector

perpendicular line in 2D
perpendicular surface in 3D

Decision plane

Key Consequences

vw<0 vw>0

Orthogonality v*w=0

Dot Product - the sum of the products of the corresponding entries of the two sequences of numbers.

Angles

Matrix multiplication and examples
noun_transform_476616

If A is a matrix where the rows are features wi and B is a matrix where the columns are data vectors vj then the i,j-th entry of the product is wi*vj, which is to say the i-th feature if the j-th vector.

In formulae: if C=AB, where A is an n x m matrix and B is a m x k matrix, then C is a n x k matrix where

Well defined:

Geometry of matrix operations

The Determinant
det(A) is the factor the area is multiplied by

det(A) is negative if it flips the plane over

Intuition from Two Dimentions

Suppose A is 2x2 matrix (mapping R^2 to itself). Any such matrix can be expressed uniquely as a stretching, followed by a skewing, followed by a rotation

Any vector can be written as a sum scalar multiple of two specific vectors

A applied to any vector

Determinant computation

Larger Matrices
m determinants of (m-1)x(m-1) matrices
computer does it simplier Q(m^3) times
called matrix factorizations

The Two-by-two
det(A)=ad-bc

Matrix invertibility

When can you invert?
it can be done only when det != 0

How to Compute the Inverse
A^(-1)*A=I

Hadamard product
An (often less useful) method of multiplying matrices is element-wise
AoB

Vectors ➡

Measures of Magnitude

Norm Properties

All distances are non-negative

Distances multiply with scalar multiplication

If I travel from A to B then B to C, that is at least as far as going from A to C (Triangular Inequality)

Types of Norms

Lp-Norm

Euclidean Norm (L2-Norm)

L1-Norm

L0-Norm
Despite the name, it's not a norm
It's a number of non-zero elements of a vector

L∞-Norm

Norms - are measures of distance

Geometry of Norms

Geometry of Column Vectors.
Operations

Subtraction as Mapping

Scalar Multiplication

Addition as Displacement

Vectors as Directions

Probability 🎲

The Gaussian curve

Standard Gaussian
(Normal Distribution) Density

E[X]=0
Var[X]=1

Key Properties

Central limit theorem

is a statistical theory states that given a sufficiently large sample size from a population with a finite level of variance, the mean of all samples from the same population will be approximately equal to the mean of the population.

Maximum entropy distribution

Amongst all continuous RV with E[X]=0, Var[X]=1. H(X) Entropy is maximized uniquely for X~N(0,1)

Gaussian is the most Randon RV with fixed mean and variance

General Gaussian Density

Visualizing Probability
using Venn diagram

Inclusion/Exclusion

Relative complement of A (left) in B (right)

Symmetric difference of two sets

Intersection of two sets

Absolute complement of A in U

Union of two sets

General Picture
Sample Space <-> Region
Outcomes <-> Points
Events <-> Subregion
Disjoint events <-> Disjoint subregions
Probability <-> Area of subregion

Chebyshev’s inequality

For any random variable X (no assumptions) at least 99 of the time

Terminology

Outcome
A single possibility from the experiment

Probability
Fraction of an experiment where an event occurs

P{E} є [0,1]

Event
Something you can observe with a yes/no answer

Capital E

Sample Space
The set of all possible outcomes

Capital Omega

Entropy

Examples

Two coins
Entropy = 2 bits of randomness

TT (1/4)

TH (1/4)

HH (1/4)

HT (1/4)

A mixed case
Entropy = 1.5 bits of randomnes =
=1/2(1 bit) + 1/2(2 bits)

TH (1/4)

TT (1/4)

H (1/2)

One coin
Entropy = one bit of randomness

T (1/2)

H (1/2)

Entropy (H)

Examine the Trees
if we flip n coins, then P=1/2^n
# coin flips = -log2(P)

The Only Choice was the Units
firstly you need to choose the base for the logarithm.
If the base is not 2, then Entropy should be divided by log2 of the base

Bayes’ rule

Posterior odds = ratio of probability of generating data * prior odds

can be leveraged to understand competing hypotheses

odds is fraction of two probabilities
i.e. 2/1

Axioms of probability

Something always happens.

The fraction of the times an event occurs is between 0 and 1.

If two events can't happen at the same time (disjoint events), then the fraction of the time that at least one of them occurs is the sum of the fraction of the time either one occurs separately.

Independence

A and B are independent if P{AnB}=P{A}*P{B}

Two events are independent if one event doesn't influence the other

Random variables

is a function X that takes in an outcome and gives a number back

Standard Deviation

Discrete X takes only at most countable many values, usually only a finite set of values

Expected Value
mean

Variance
how close to the mean are samples

Continuous random variables

For many applications ML works with continuous random variables (measurement with real numbers).

Probability density function

Building machine learning models

Maximum Likelihood Estimation

Given a probability model with some vector of parameters (Theta) and observed data D, the best fitting model is the one that maximizes the probability

Conditional probability

If I know B occurred, the probability that A occurred is the fraction of the area of B which is occupied by A

Intuition:
The probability of an event is the expected fraction of time that the outcome would occur with repeated experiments.

Univariate Derivatives 👤

Newthon's method

Idea

Minimizing f <---> f '(x)=0

Now look for an algorithm to find the zero of some function g(x)

Apply this algorithm to f '(x)

Update Step for Minimization

To minimize f, we want to find where f '(x)=0 and thus we may start with some initial guess x0 and then iterate Newton's Method on f ' to get

Update Step for Zero Finding

we want to find where g(x)=0 and we start with some initial guess x0 and then iterate

Relationship to Gradient Descent
A learning rate is adaptive to f(x)

Pictorially
g(x) x such that g(x)=0

Computing the Line

line: on (x 0, g(x 0))
slope g '(x 0)
y=g '(x 0) (x-x 0)+g(x 0)
solve the equation y=0

Maximum Likelihood Estimation

find p such that Pp(D) is maximized

Gradient Descent

ALGORITHM

1. Start with a guess of X0

2. Iterate through
- is learning rate

3. Stop after some condition is met

if the value if x doesn't change more than 0.001
a fixed number of steps
fancier things TBD

Issues:

Sometimes we can circumvent this issue.

recall that an improperly chosen learning rate will cause the entire optimization procedure to either fail or operate too slowly to be of practical use.

how to pick eta

As simplistic as this is, almost all machine learning you have heard of use some version of this in the learning process

Goal: Minimize f(x)

Derivative

Rules

Chain Rule

Alternative

Quotient Rule

Product Rule

Sum Rule

Can be presented as:

Interpretation

Most usable

http://hyperphysics.phy-astr.gsu.edu/hbase/Math/derfunc.html

let's approximate

better approximation

f(x +є) = f(x) + f'(x)є

Second Derivative

f''(x)
shows how the slope is changing

max -> f '' < 0
min -> f '' > 0
Can't tell -> f '' = 0
proceed with higher derivative

Multivariate Derivatives 👥

Convexity

the function is convex if the line between two points stays above

Derivative Condition
Hf is called positive semi-definite if

And this implies f is convex

A Warning
When going to more complex models i.e. Neural Networks, there are many local minima & many saddle points.
So they are not convex

Benefits
When a function is convex then there is

a single unique local minimum
no maxima
no saddle points
Gradient descent is guarantied to find the global minimum with small enough
Newton's Method always works

Opposite is Concave

Matrix Calculus

The Gradient

Key Properties

points in the direction of maximum increase

- points in the direction of maximum decrease

at local max & min

Definitions

The Gradient
collection of partial derivatives

Vector derivative

Matrix derivative

Visualize

Gradient Descent

Level set

Newton Method

The computational complexity of inverting an n x n matrix is not actually known, but the best-known algorithm is O(n^2.373)

For high dimensional data sets, anything past linear time in the dimensions is often impractical, so Newton's Method is reserved for a few hundred dimentions at most.

Second Derivative

Hessian

2D intuition

Critical points
=0

det(Hf) > 0
further investigation

tr(Hf) = 0
does not happen

tr(Hf) < 0
local maximum

tr(Hf) > 0
local minimum

det(Hf) < 0
saddle point

det(Hf) = 0
unclear
need more info

Hf=

If the matrix is diagonal, a positive entry is a direction where it curves up, and a negative entry is a direction where it curves down

For , there is many derivatives

Trace
sum of diagonal terms tr(Hf)

Partial Derivatives
is a measure of the rate of change of the function... when one of the variables is subjected to a small change but the others are kept constant.
Example:

If you have a function
where n - n-dimensional vector,
Then it's a function of many variables.

You need to know how the function responds to changes in all of them.

The majority of this will be just bookkeeping, but will be terribly messy bookkeeping.

UX Design

GUI Graphical User Interface

Ressources

The Encyclopedia of Human-Computer Interaction, 2nd Ed.

15.2 From Usability to User Experience - Tensions and Methods

15.2.3 The Long and Winding Road: Usability's Journey from Then to Now

15.2.1 New Methods, Damaged Merchandise and a Chilling Fact

15.2.4 Usability Futures: From Understanding Tensions to Resolving Them

15.1 From First World Oppression to Third World Empowerment

Usability Evaluation

the extent to which an interactive system is easy and pleasant to use

evaluation

designing requires making choices.

Evaluations have to be designed

choose how to evaluate.

propositions

evaluation methods

measurable property

Evaluation methods and metrics are thoroughly documented in the Human-Computer Interaction research and practitioner literature.

15.1.1 The Origins of HCI and Usability

time

1960 -> minicomputers

1980 -> no / basic training on operating systems and application

1946 -> born of computer -> specialists of expensive centralised equipment

problem

HCI was born

software design practice -> assume knowledeable and competent users

15.1.2 From Usability to User Experience via Quality in Use

user experience

usability

how inherently usable an interactive system was -> how well it fitted its context of use.

Quality in use

15.1.4 From HCI's sole concern to an enduring important factor in user experience

15.1.3 From Trouble with Computers to Trouble from Digital Technologies

1999 -> hole in the wall -> children learn computers

15.3 Locating Usability within Software: Guidelines, Heuristics, Patterns and ISO 9126

15.3.2 Manageable Guidance: Design Heuristics for Usable User Interfaces

15.3.3 Invincible Intrinsics: Patterns and Standards Keep Usability Essential

15.3.1 Guidelines for Usable User Interfaces

15.4 Locating Usability within Interaction: Contexts of Use and ISO Standards

15.4.1 Contextual Coverage Brings Complex Design Agendas

15.5 The Development of Usability Evaluation: Testing, Modelling and Inspection

Design of HCI

task analysis

HTA

stopping rules

hierachis of goals and subgoals

tasks

representation of analysis

Goals

GOMS

goals

methods

selection rules

operations

levels of analysis

micro

cognition

intermediate

task

macro

environment

interface surveys

labelling surveys

enviromental surveys

walkthrough approaches

cognitive walkthrough

data collection

verbal protocols

questionaries

structured interviews

Ergonomic checklists

user analysis

user population

interesting characterisics

age

education

IT knowledge

sex

physical limitations

personae

not real person

very detailed

requirements gathering

indirect observation

direct obsveration

stakeholders

e.g. managers

interface specification

includes

layout

state changes

usage scenarios

error messages

behaviour

description of interaction

description of physical interface

Specification Vs Prototyping

prototyping

represent

system

task

user

types

paper

low fidelity

interaction not real

easy to set up

full

rarely used

lots of time coding

horizontal

longer to build

give users scope for making mistakes

slide show

limited representation

good simulation

e.g. visual basic

high fidelity

vertical

one part of system fully implemented

HCI software process

https://coggle.it/diagram/XbOhdK4hAcR1viaq/t/the-encyclopedia-of-human-computer-interaction%2C-2nd-ed

https://coggle.it/diagram/XbOiWnoA8nxvssP3/t/hci-revision

15.2.2 We Can Work it Out: Putting Evaluation Methods in their (Work) Place

https://coggle.it/diagram/WlsLPFN30QABoeQB/t/machine-learning-approaches

https://coggle.it/diagram/Wf5mYoJbsgABUF9P/t/neural-net-arch-genealogy

Ch 2 - Understanding and conceptualising interaction

Paradigms, visions, theories, models, frameworks

Frameworks

Norman's framework

User

Designer

How the designer thinks the system should work

System

The system image is how the interface, manuals, etc communicate the workings of the system to the user

Numerous frameworks exist

Variety of forms - steps, questions, concepts, challenges, principles, tactics, dimensions

Offer advice to designers - what to design or look for

Traditionally based on theories of human behaviour, increasingly derived from actual design experience and user studies findings

Theories

Cognitive

Social

Organisational

Different theories have entered the HCI realm over the last 30 years, as a way to analyse and predict the performance of users of computer base interfaces

Paradigms

How findings should be analysed and interpreted

The phenomena to be observed

The questions to be asked and how they should be framed

Augmentation (late 1990s) - sensors + algorithms in/on homes, public buildings, bodies - parking lots which tell you where the open bays are, how well you slept and whether your blood pressure is too high / Internet of Things

Ubiquitous tech (Weiser 1991) - tech embedded in everyday objects, in an unobtrusive way

1980s - user centred applications for desktops / screens; WIMP = Windows, Icons, Menus, Pointer

Visions

Visions of the future drive research and design

The inventor of the mobile phone was inspired by Star Trek, tablet like devices were seen in 2001: A Space Odyssey (made in 1968)

Models

They are usually derived from a theory from a contributing field such as psychology

Norman's models of interaction design (e.g. 7 stages of action model, model of emotional design) were based on cognitive science

Keystroke model (Card, Moran and Newell) was also based on cognitive science

Conceptual models

Working strategy and framework of general concepts and their interrelations

Concepts that people are exposed to through the product including task-domain objects

Attributes e.g. name, description, date added

Operations that can be performed on them, e.g. add, update, delete, move, organise

Mappings between concepts and user experience e.g. having easy access to useful sites, being able to group them in your own categories so that they are easy to find, being able rename them so that they are easy to recognise

Relationships e.g. categories, groups, tags

Metaphors and analogies that enable people to understand what the product is and how to use (e.g. bookmarking)

Johnson and Henderson's definition

"Enables designers to straighten out their thinking before they start laying out their widgets"

"High level description of how a system is organised and operates"

Examples of conceptual models and interface metaphors

Operating systems use the physical office idea of "desktops", "folders", "files", "trashcans" metaphors

Task management tools use "post-its" metaphors

Online shopping sites use the physical shopping mall concept with "stores", "shopping carts", "checkout" metaphors

E-readers use "bookshelves" metaphors

Google's Material Design uses the paper metaphor

Problem space

Team effort

Operational impact

Financial implications / budget

Technical solutions / impact

Customer impact

Overview

Understand how this can be improved

Understand the current user experience

Is there a problem to be solved?

(Don't just start designing)

Research first

Framework

Could your design create new problems? (not from the textbook)

How does the proposed design solve these problems?

What is it?

Why do you believe there is a problem?

If there is no problem but you are trying to improve the UX, how do you believe your design achieves this?

Is there a problem?

Benefits

Open mindedness - prevent the design from becoming narrowly focused too early

Common ground - establish a common language to reduce confusion and misunderstandings

Orientation - ask specific questions about how the conceptual will be understood by users

Assumptions and claims

Claim: We only need to support smartphones. Assumption: All of our customers surely have smartphones by now. Reality: 20% of our customers still use feature phones.

Claim: We need to redesign our search tool. Assumption: Our search tool is really unusable and that's why nobody uses it.
Reality: The type of content on our site is not well suited to search.

Interaction types

Exploring

Moving through a physical environment (with sensors), e.g. Apartheid Museum, or a #D virtual environment e.g. Minecraft, Subway Surfer

Manipulating

Core principles for manipulation

Continuous representation of objects and actions of interest

Physical actions / pressing buttons instead of issung complex commands

Rapid reversible incremental actionawith immediate feedback

Moving, opening/closing, zooming, dragging items using a physical device (.e.g mouse, Wii) or gestures, or fingers against a touch screen

Conversing

Spoken or typed dialogue, e.g. search using voice input or Web browsers for the speech impaired

Instructing

E.g. print a file (PC), give me a coldrink (vending machine), delete a photo (phone)

MICE

Interface metaphors

Provide familiar entities that enable people to understand the underlying conceptual model, but do not usually function in exactly the same way, which can confuse people

Shopping cart - I put items I want to buy in there, but in the real world I have to buy them right away or the cart is emptied; if I get books from the library I don't put them into a cart

Trashcan - for getting rid of unwanted stuff - but why is it on the desktop?

Folders - I can put related items into a folder, but in the real world there is a limit to how many items

How the user thinks the system works

Virtual Reality

Applications of VR

Data visualization, science, archaeology

Industrial design

Entertainment and art

Health and wellbeing

Psychological research and mental health therapy

The MIND project

Used at the clinic, as part of a regular 'face-to-face' consultation

Experience the mind's dynamics, learn powerful therapy tools and manipulate mental processes with your own hands

Modular design allows to build a suite of apps over time

History

Started since 1950s: Sensorama

Current VR apps: games, simulation

Gartner Hype cycle for emerging technologies: innovation trigger -> peak of inflated expectations -> trough of disillusionment -> slope of enlightenment -> plateau of productivity

Current day: mobile VR headsets, wired VR headsets

Introduction

3-D audio

Inputs and outputs: computer animation is the same as desktop games, but the camera position is controlled by the player (translation and rotation of camera is affected by the player's head movements)

An embodied sensory experience of a synthetic environment

Hand controller: to move objects in the virtual space

Head mounted display: Need to track player's head movements to change position of camera in the virtual space

Conditions addressed by VR in mental health

Anxiety (bulk)

Psychosis

Depression (almost nothing)

Enabling exposure therapy, Thought & behaviour therapy, to be more widely used

https://coggle.it/diagram/XbOoBHoA8pP5ssc9/t/18-virtual-reality

Author: https://twitter.com/tllhglld

State: Draft (not finished yet)

https://coggle.it/diagram/XbOer64hAVFRviPP/t/math-for-machine-learning

Licence: CC

Big Data

Open Data

other maths

Workload management

GPU / CPU

Slurm

Google Collab

Embedded Systems

Lateral "Side"

Datasets

Commands

Job Management

Accounting

Job Submission

Daemons

Environement Variables

https://slurm.schedmd.com/pdfs/summary.pdf

squeue

sinfo

scontrol

scancel

salloc

sbatch

srun

IOT

NLP

Natural language Processing

Open Source Libraries

Apache OpenNLP

Mallet

Natural Language Toolkit

Standford NLP

Terminology

Morphology

Phonology

Pragmatic

Discourage

Semantic

Morpheme

Syntax

World Knowledge

History

1970s

Conceptual Ontologies

1950s

Turing Test and George Town Experiment

1980s

Machine learning algorithm

1960s

Elizabeth pyschotherapist

Areas

Named entity recognition

Part of speech

Optical charaacter recognition

Applications

Extracting data from set

MachineTranslation

Database Access

Information Retrieval

Text Categorization

Components

Natural Language Understanding

deals with machine reading comprehension

Natural Language Generation

a translator that converts data into a natural language representation

Introduction

The field that covers computer understanding and manipulation of human language, and it’s ripe with possibilities for newsgathering

The way for computers to analyze, understand, and derive meaning from human language in a smart and useful way

The field of study that focuses on the interactions between human language and computers

Steps

Syntactic Analysis concerns the proper ordering of words and its affect on meaning

Pragmatic Analysis concerns the overall communicative and social context and its effect on interpretation

Lexical analysis involves dividing a text into paragraphs, words and the sentences

Semantic Analysis concerns the (literal) meaning of words, phrases, and sentences

Discourage Integration concerns the sense of context

https://coggle.it/diagram/XbQngkUb1ndwn-sJ/t/natural-language-processing

Cybernetics

Engineering / Libraries for ML

Data Mining

Anterior "Front"

Neuropsychology

Executive Functions

Ressources

Neuroscience: Historical Perspectives

Localisation and neuroanatomy

17th Century: Cartesian Legacy

Descartes: brain and behaviour mediated by a mechanical theory that nerves are hollow with fluid flowing through them - 'Balloonist' Theory. Believed mind located in pineal gland. Dualism mind and body are separate!

Pineal gland: small endocrine gland in brain between two hemispheres (central so logical to place mind here), it produces melatonin.

Willis: post-mortem examinations, concluded that psychological attributes are heavily reliant on the cortex! Dualism but believed the point of contact between mind and body is the cortex. ⛔Cartesian Gap

16th century: Non-Ventricularist

rejected previous theory! Evidence that all mammals have the same anatomical organisation as humans but do not have the same intellect. Believed that rather than the soul, it was animal spirits which reside in ventricle and follow nerves to organs.

13th Century: the mind and brain are related but seperate

Magnus brain is the centre of mental activity: sensation, rationing, memory. Mind and brain inextricably linked, but mind immaterial and resides in ventricles.

19th century: Advances in microscopy, improved histology, debate about nervous system organisation, Fritsch & Hitzig

15th century: the Rebirth of Neuroanatomy!

Da Vinci: anatomical contribution of structure of the ventricles. Believed soul resides in ventricles rather than tissue because it is incorporeal.

18th century was useless!

The Neuron Doctrine

Ramon y Cajal neuroanatomist: The Neuron Doctrine!: acceptance of the neuron as anatomical and functional unit of the nervous system

Investigated nervous conduction: direction of impulses and idea of synapses.

Won Nobel Prize (against Golgi!)

1887: new staining method! fixative and silver nitrate twice to get deeper staining... lead to observation in chicken cerebellum samples that dendrites reside at axon endings.

"each cell is a totally autonomous physiological canton" Cell theory

"the relationship between nerve cells was not one of continuity but rather of contiguity" --> not continuous, just in close proximity

Golgi staining the silver nitrate method invented by Golgi in 1873

won Nobel Prize (against Cajal!)

"Rete Nervose Diffusa" in Italian which literally translates to widespread nerve network

Golgi believed his observations of cerebellar tissue etc were evidence of a diffuse nerve network hence supporting reticulum approach - as he observed continuous network rather than discrete nerves.

a more holistic approach

Reticularists v Cell Theorists

Both: use the same methods but came to different conclusions ⚠low magnification and poor resolution of microscopes available at the time

R: believe NS is a large network of tissue (reticulum), formed by fused nerve cells. "The whole brain is a syncytium" meaning a large, cell-like structure. Von Gerlach

C: believe that NS consists of distinct nerve cells. Controversial theory that autonomous life should be attributed to cells. Virchow "omnis cellula e cellula" cells can only multiply from cells.

Reasons for delay: why did it take so long to realise this?

belief of cell continuity was necessary for cell interaction at that point of theoretical knowledge: i.e. did not make sense to scientists that discrete units could interact- they must be connected!

Technical advancement of histological prep and microscope accuracy

nerve cells/ NS have very complex structure!

1950s: Electron microscopy techniques: Bennett: Reaffirmed Cajal's proposal that there is a gap (synapse) between all nerve endings, even when greatly intimate.

advancing technology aided advancement in understanding of the nervous system and individual nerve cells.

Dendrites, cell body, axons identified.

Main Tenets of the theory

information flows in one direction along the neuron

Neurons are discrete not continuous

The fundamental structure/ unit of the NS is the neuron

Neurons have 3 parts: dendrite, cell body, axon

1820 Purkinje cells: Johannes Evangelista Purkinje: discovery of these cells made him famous, they are among the largest cells in the vertebrate brain

Waldeyer: coined the term 'neuron' in 1891

Early theories: the mind and body

Galen of Pergamon: Encephalocentrism: anatomical methodology, brain receives sensations and generates understanding.
"apodeictic proofs" provided: brain and spinal cord source nerves. Spinal cord originates from brain

Experiments: severing throat nerves in pig inhibited squealing (but still breathing), concluded that voice comes from brain and not the heart.

Soul: communicates with the body via spirits, through 3 ventricles (brain cavities!) & travel through body via nerves.

Plato: the brain conveys sensations and once these settle in the mind, knowledge, memory and opinions are produced.

Encephalocentrism but also cardiocentrism!: 3 species of soul

Epithemetikon: perishable soul, responsible for urges e.g. hunger, passion, located in liver

Thymos: perishable soul, responsible for emotions, located in chest/heart

Logos: immortal, divine, localised in head

Western Medicine: Hippocrates: noted 2 hemispheres of the brain; 'lateralization' and rejected possibility of a 'sacred' cause of epilepsy: said it was a disease of the brain. Believed the brain is the most powerful organ

Encephalocentrism vs Cardiocentrism debate

Where lies the 'seat of human consciousness, sensation and knowledge'? Brain vs heart!

Aristotle: 3 soul faculties, all reside in heart: cardiocentrism

Nous: humans, an immaterial soul

Vegetative: plants and some animals

Believed the role of the brain to be: diminish heat of blood generated by heart and generate sleep

Sensitive/motor: animals

Debate continued into the Renaissance!

Homer: One overall 'soul' called 'psyche': non-localised soul which represents individual life/identity. soul is active/non-silent only in dreams and abandons the body at death

three additional types of 'body soul' located in chest

Noos: reasoning

Menos: aggression

Thymos: emotion

Egypt! 48 cases of head injuries investigated: noted the importance of the brain and it's role in behaviour but believed the heart was the 'seat of the soul' (mummification)

Natural Philosophers: systematically reject supernatural and try to understand everything through material mechanisms

e.g. Anaximenes who theorised that the origin of human thoughts is air.

Anatomical dissection: Alcmaeon of Croton: Encaphalocentrism: sensory and cognitive significance of brain lead him to believe it was where consciousness resides & if the brain is disrupted then so are all senses. Also, believed that man has understanding whilst animals only have sensations

Cardiocentrism: THE HEART IS...: affected by emtion, all animals have one, source of blood, warm, connected to whole body, essential, formed first in foetus, sensitive to pain, central in body!

Localisation: Phrenology to cortical topography

"the first brain map": Fritsch & Hitzig: dogs' cortex electrically stimulated resulted in muscular contractions. ✅ localization, as findings generalised to say that all functions are linked to specific areas.

19th century

Localization theory: high degree of specialization of brain areas' functionality Gall & Spurzheim

Brain Equipotentiality theory: all parts of the brain have equal significance in functions

1585: von Grafenberg: early observations of brain damage - although the tongue was not paralysed the patient could not speak: Same as Galen - voice must come from the brain

Anatomy & brain mapping: Brodmann: systematic investigation of primates developed somatosensory map: adjacent areas of body represented in adjacent areas of cortex.

Jackson: Epilepsy: often convulsions are on one side of body only - perhaps because the organic disease is only affecting one hemisphere?... motor cortex organised somatotopically: i.e. hands are most sensitive/wide range of movement so must receive largest representation in cortex.

more research builds bigger picture: Penfield

Homonculus!

Phrenology: Principles: Many organs in the brain, responsible for particular skills. Bigger = more powerful. The brain shape is determined by organ development. Skull surface provides accurate index of brain. Can analyse personality using this method. Phrenometer: machine which measures 'bumps' on skull.

but they (Gall & Spurzheim) did make important observations, such as the fact that the brain is folded.

in reality, skull shape has nothing to do with cognition.

Brain mapping: lesions: Broca's area & aphasia: left inferior frontal cortex damage thought to be cause of language/speech inhibition.

Against localization: Flourens: brain lesions in animals did not cause specific behavioural deficits, hence concluded that brain shares functionality: Aggregate field theory

Mapping and understanding of 'mind': explain beh in mechanistic terms, dualism -> materialism, reductionism: explain brain in terms of its parts ❌ does this really inform cognitive models? are mind and brain separate?

Wernicke's aphasia: damage to posterior left hemisphere: language understanding rather than speech.

https://coggle.it/diagram/XbQvj0Ub1g6An_AA/t/neuroscience-historical-perspectives

⛔Ghost in the machine: Cartesian Gap how could something immaterial influence something material?

Neuroscience:Research Contribution to CS

Cognitive Neuroscience: The Biology of the Mind

Human Brain: The 3 Main Parts

Cerebrum

Its outer folded layer is called
the cerebral cortex.
 The surface of the cerebrum is
the cerebral cortex

Cerebral Cortex

FOUR MAJOR AREAS

F rontal

Higher
functionslanguage,
thought,
memory, motor
functions,
problem solving,
movement

Studies have shown that the frontal area is the most
common region of injury following mild to moderate
traumatic brain injury

Extremely vulnerable to injury
due to their location at the front of the cranium

P areital

Receive signals from the touch system,
important for vision and attention,
perception of stimuli

Receives info. from the
senses about pressure,
texture, temperature,
and pain

T emporal

Language, memory, hearing, vision

Primary function:
 Hearing
 Or auditory processing

O ccipital

First place in
cortex where
visual
information is
processed

Primary function
 Vision
 Or visual processing

 3 mm thick
 Very thin layer
 The area is a convoluted pattern
packed into a small space inside the
skull.

Surface of brain = cortex
(i.e. gray matter)

Decision making organ of
the body
 Receives messages from
all sensory organs
 Initiates all voluntary
actions
 Stores all our memories
 Where our knowledge of
language resides

Cerebral Hemispheres

Composed of cerebral
hemispheres (left & right),
joined by corpus callosum

Corpus Callosum = large band of fibers that connects the
left and right hemispheres of the cerebral cortex

Contralateral brain
function
 Left hemisphere controls right
side of body
 Right hemisphere controls left
side of body

The left and right hemisphere
are separate, but connected

The cerebrum is the largest
part of the brain and controls
all conscious thoughts,
experiences, and actions

Divided into right and left
hemispheres, which are joined
by the corpus callosum

Brain Stem

It controls processes basic for
survival, such as heart rate,
breathing, digestion, and
sleep.

It also has its own set of nerves
that send and receive signals to
the face, mouth, tongue, eye
muscles, ears, and balancesensing
vestibular organs

It is the main route of
communication between the
rest of the brain, the spinal
cord, and the nerves that run
throughout the body

Cerebellum

Recent research also
suggests a role in higher
cognitive processes

The cerebellum is the
second largest part of the
brain

It controls posture and
balance.

Human Brain

Approximate length = 15 cm

Consists of approximately 10
billion nerve cells (neurons) and
billions of interconnecting fibers

Approximate weight = 1400 g ( 3
pounds/ 1.4 kg)
 Newborn = 350 to 400 g

Most complex organ of the body

Brain: The Nervous System (Vertebrate)

Neurons

Dendrite = Receives messages.
Brings information to the cell
body

Neuron connections

This makes the brain the most
complex organ and most complex
structure in this universe

Huge number of interacting
neurons in the brain

Axon = Sends message. Takes
information away from the cell
body

Soma: body of the cell.
Contains genetic material

nerve cells, specialized to receive and transmit
information in the nervous system
 Communicate through stimulation of electric impulses/signals

Synapse = a small gap
separating 2 neurons
 a link from one neuron to
another

Neurotransmitters

This makes our brain work
 determine how we feel, think and
act

Chemical signals that are released
across the synapse so that neurons
can communicate with each other

Brain Research: The Development

Insights from Psychology: Sheperd Ivory Franz

Training of animals in new habit after removal of brain
parts (frontal lobe)
 New habits lost but old ones retained
 Animals could relearn lost new habits

Generalized to human
 Functional restitution or reeducation after brain lesions –
evidence showed that lost function can be recovered (i.e.
functions could not have resided in damaged area alone)
 Brain = integrated whole

Challenged theory of localization of function based on
findings on study of experimental animals’ behavior

Late 19th & Early 20th Century

Brain – the organ of the mind

Overall – primary investigative techniques
employed include electrical stimulation or
removal of specific brain areas
 Theory of localization of function

David Ferrier continued experiments on other
species (i.e. monkeys, frogs, guinea pigs) to find
out whether the theory can be generalized
 Used the method of ablation (surgical removal) or
also called localized destruction
 Extended investigations beyond “motor centers” to
include sensory centers
 E.g., vision (perception) and hearing

Insights from Psychology: Karl Lashley

Proposed Principles

Principle of Equipotentiality

 Parts of the brain have the
equal potential to fulfill the
function of a damaged part
 The intact part of the brain
can carry out the functions
that were lost in the
damaged part. This capacity
varies from one area to
another and from one
function to another.

Principle of Mass Action

When a functional area of
cortex is damaged, its ability to
perform previously mastered
complex functions may be lost
in proportion to the extent
of the injury

Proposed 2 principles to explain his conclusion
 Principle of Equipotentiality
 Principle of Mass Action

Theory of localization – oversimplified; argued instead that
functions of every center dependent on its relation to rest
of intact nervous system
 Based on findings from experiments with white rats which
investigated their learning & intelligence
 Intelligence = global function because learning can recur
(relearn)

Late 19th & Early 20th Century

Evidence in support of the theory based on
Fritsch & Hitzig’s experiments on dogs
 Muscle movement stimulated by direct electricity
application to brain

Hughlings-Jackson’s observation of recurrent
patterns of epileptic seizures in patients -
connection between stimulus event in
specific brain areas and muscular response of
specific body side
 Localization of function in the two
hemispheres
 Theory of contralateral control (opposite-side)

Late 18th & Early 19th Century

Gall
 Conclusion that structure & function were related
 Structure of brain & people’s behavior
 Phrenology
 Emergence of theory of localization
 Broca’s Area
 Wernicke’s Area

The Neuroscience Approach (Cognitive Neuroscience)

Neuroscientists

Learn more the structures and functions
of the human brain and how they relate
to the behaviors we observe in people:
 using language, solving problems, and
remembering

Attempt to explain cognitive
processes in terms of underlying
brain mechanisms

Describe the biological ‘hardware’ upon
which mental ‘software’ supposedly runs

Brain Plasticity

Radical suggestion, radical solution

Hemispherectomy

The brains of children have amazing abilities to
rewire themselves

Since the left side of the body is controlled by the right side of the brain,
the patient would be paralyzed on the left side when
they are awoke

Removal of half of the brain

Surgery to have one side of the brain removed to prevent
seizures.

Someone could be playing and become suddenly rigid
and collapse on the floor

Phantom Limbs

Approximately 5-10% of individuals with an amputation
experience phantom sensations in their amputated limb, and the
majority of the sensations are painful

Sensation that an amputated or missing limb (even an organ, like
the appendix) is still attached to the body and is moving
appropriately with other body parts

The Living Human Brain: The
Neurosurgeons

Insights from Wilder Penfield (Neural Cartographer)

Construct MAP of brain areas
 Localization of various functions
that displayed:
 Muscular contractions
 Tingling sensations in selected body parts
 Mental phenomena – Hearing Music
 Memory - Recall of past experience
 Psychical states – feelings of fear, loneliness

Neuroscience: Modern Technology &
Research

Evidence from Brain Mapping Research: Insights into Human Cognition

Disorders

Schizophrenia- severe mental disorder (hallucinations
& delusions)

 Structural image differences between patient with
childhood onset schizophrenia & adult
 Medication has effect in patients’ brain (caudate –
basal ganglia)
 Language dysfunction due to deficit in language
processing ability (unable to integrate context –
incoherence)

Bipolar &
Unipolar
Disorder

 Mood – Bipolar (normal mood alternated with both depression
& mania) & Unipolar (only depression)
 Role of neurotransmitter - Missing or deficient
chemical level or substances
 Neurochemical and autonomic abnormalities.

Autism

 developmental brain disorder affecting brain function –
affect social interaction (social & communication deficits)
 Greater brain volume found in autistic subjects – possible
causes
 Increased production of neurons
 Neurons do not die off in great numbers as in
normal development
 Increased production of non-neural brain
tissue (e.g. blood vessel)

Dyslexia

 Brain-based type of learning disability – affect reading
 Unawareness of associations between phonological
segments (sounds) in construction of words
 Inability to decode

Language

Aphasia
(ASL)

Localization
of Brain

Brain
Plasticity

Processing of
Content &
Function Words

 Neural organization of language quite similar for
language learned in ordinary way (i.e. hearing &
speaking) and for signed language acquired by deaf
individuals
 Insight from fMRI images on Broca’s & Wernicke’s
Areas confirm hypotheses about involvement of
these areas in human language function
 Left hemisphere’s involvement in language is the
same, whether the language is seen or heard
 The intact hemisphere can take over tasks of
damaged/removed brain hemisphere
 Different parts of brain involved in processing
content (nouns) & function words

Concept
Categories

Categorization
Task

Name finding

Memory

Functions of
Working
Memory

Role of
Hippocampus

Implicit &
Explicit

Memory
Long Term
Memory

 Two distinct functions of working memory (i.e. retrieval &
coding) are performed in different locations within the
brain area associated with working memory
 working memory NOT generalized; different sorts of
processes carried out in different regions
 Investigate role in explicit (i.e. conscious recollection) &
implicit (i.e. unconscious recollection) memory of events
 Activation associated more with actual/explicit recollection of
event (i.e. conscious) rather than in effort trying to remember it
 Significant role of prefrontal cortex in explicit memory
 Greater activation at higher levels of load (higher
memory load lasts longer - continuing activation)
 Amount of memory load = amount of information to be
remembered

Brain Mapping Techniques

Functional

Blood-Flow Techniques
 Positron Emission Tomography
(PET)
 Single Photon Emission
Computed Tomography (SPECT)
 Functional Magnetic Resonance
Imaging (fMRI)

Positron Emission Tomography (PET scan)

Disadvantages

 Requires the use of cyclotron,
an expensive equipment to
provide the radioactive isotopes
(decay easily and needs to be
produced each day)
 Not as sensitive to changes over
milliseconds (in comparison to
electrophysiological techniques)
 not as precise for analyzing
cognitive tasks involving
changes within milliseconds
 Radioactive material used

Method of measuring
cerebral blood flow while
subject carries out cognitive
tasks
 Uses radioactive isotopes
(i.e. positrons)

Advantages

Provides an image of brain
activity as a cognitive task is
occurring – locate activity in
specific parts of the brain

PET scans look at bodily
process by detecting the
decay products from
radioactive tracers injected
into the body.

Functional Magnetic Resonance Imaging (fMRI)

Advantages

 Provides information about
brain structure and
function
 Safer- Less harm to patients
as no X-rays or radioactive
material is used
 Allows researchers to infer
which locations are involved
in specific activities

 Detect changes in magnetic
state of blood using MRI
scanners with fast imaging
techniques
 Record changes in oxygen
level & blood flow in various
brain locations as subjects
perform various cognitive tasks

FMRI is a technique for determining which parts of the
brain are activated by different types of physical sensation
or activity, such as:
 sight
 sound
 movement of a subject's fingers
 mental imagery
 calculation

Also known as real time or
dynamic MRI – available in
1990s
 Decade of the Brain

Electrophysiological Techniques
 Electroencephalography
 Event-Related Potentials (ERPs)

Electroencephalograms (EEGs)

Provide fine
-tuned reading
of rapidly occurring
changes (MRI too slow to
do this)

Tracings/Recordings of
patterns of electrical
activity in the brain  view and record the changes
in brain activity during
performance of cognitive
task

Electroencephalography

Disadvantages

 Invasiveness – relative
 May affect data validity
 Possible interference from
movement, heartbeats –
blur reading

A test of the function of the brain itself
 records the electrical activity on the brain's
surface.
 image the brain while it is performing
cognitive task.
 detect the location and magnitude of brain
activity involved in the various types of
cognitive functions

Advantages

 Able to trace rapid changes
in neural activity
 Can record activations in
brains of people who are
fully conscious & engaged
in various activities in
natural environment

Non-invasive and painless procedure
 Take brief patient history
 Apply electrode leads to the patient's scalp
 Run the test

Event-Related Potential (ERP)

Evoked response provides a
picture of neural activity
changing over time as the
brain processes information

 Subject presented with a
stimulus during EEG process
 Record voltage change around
the stimulus (before and after
it ends)
 Repetitions of event – average
the values

Structural

Magnetic Resonance Imaging
(MRI)

Advantages

 No X-rays or radioactive
material is used (unlike
CAT or PET)
 Resolution of the image is
sharper / clearer than
CAT scans
 Flexibility – allows the
researcher to distinguish
different structures of the
brain – gray/white matter
and cerebrospinal fluid

Uses a magnetic field to take
images of the inside of your body.
 Assumes atoms in the body will
react to magnetic field
 Based on radio signals emitted by
the protons in the human body

Disadvantages

 Slow – for an image to be
generated.
 An image generated at a given
time will no longer depict a
situation precisely as it was at the
time of the imaging
 Cannot be used in patients with
metallic devices, like
pacemakers
 May cause claustrophobic
reaction in some patients

X-ray Techniques
 Cerebral Angiography
 Computed Tomography
 Computer Axial Tomography
(CAT)

Computer Axial Tomography (CAT)

Disadvantage

 Doesn’t indicate when an
activity is occurring in the brain

Developed in early 1970s
 Less radiation than traditional x-ray.
 Computer technology enable
‘deblurring’ of picture &
reconstruction of a complete brain
image from multiple views
 Development of CAT enable
process of imaging brain internal
structure to become safer & more
precise– 3D

Advantages

 Safer - The fan shaped beam
exposes the body to less
radiation than traditional xrays
 More precise - only takes a
cross section of an organ,
avoiding problem of
interference of all the layers
of tissue present in x-ray
images
 Clearer image of brain
structures – ‘deblurring’ by
computer technology – 3D

Useful in detecting disorders associated with
abnormalities in brain arteries
 Example: hemorrhage (heavy bleeding)
 Loss of oxygen in that area

A procedure that uses a special dye
(contrast material) and x-rays to see how
blood flows through the brain.
 Injection of a dye into the vertebral artery
or carotid artery in the neck carried to
brain arteries
 Paths of arteries surrounding brain tissue
 X-rays of skulls then locate malfunctioning
artery – damaged area

Brain Mapping

Attempt to provide a
complete picture about
how the brain works

Relate the brain's
structure to its
function

Find what parts of the
brain give us certain
abilities.

Continued brain exploration based on opportunities
presented while treating patients (i.e. epileptic seizures,
brain tumors)

Brain Research from Neurosurgeons

Discovered that stimulations could also trigger
sensation, without muscular movements

Sought to investigate ‘the pulsation of the brain’
 Procedure conducted while the patient was still
conscious
 inserted the electrode (with weak electric current)
in the human brain
 Findings = stimulating localized brain areas could
produce specific muscular contractions
 Conscious patients could provide information during
procedure

Insights from Wilder Penfield

 Surgeon + experimenter
 Professor of neurology
 400 operations on the brain.
 Patients with epileptic seizures and brain tumors.

Insights from Donald O. Hebb

Attempted to explain how neural circuits work
(i.e. underpins cognitive processes)
 Any two cells or systems of cells that are repeatedly active at the
same time will tend to become ‘associated’, so that one activity
facilitates the other
 Cell Assembly

Brain Research: 1960s & Beyond

The visual system

 David Hubel
 Torsten Wiesel

 Connections between the eye and cortical neurons are
permanently disrupted when kitten is deprived of the
use of one eye for a period, by closing its eyelid.
 That particular brain area becomes impeded.

 Visual field = an area that is visible to you while
your eyes are not moving
 Area of the visual field that will produce a response
in a given cell = receptive field
 Hubel & Wiesel found neurons that specially detect
certain features => feature detectors
 E.g.: optic nerves of frogs have “bug detectors.”
 Connections to neuron impeded when the use of
the organ (i.e. eye) is also impeded

Split-Brain Research

Roger Sperry

 Research that includes severing of the corpus
callosum

 Research by removing the corpus callosum was originally
done on monkeys.
 After surgery, the monkeys behaved normally.
 Procedure was then employed with human patients who
suffered from epileptic (sawan) seizures

 Investigated the role of the two hemispheres
 In a normal brain, stimuli entering one hemisphere
is rapidly communicated by way of the corpus
callosum to the other hemisphere
 Brain functions as a unit

Specialization of the hemispheres

Right side
 Recognition of faces
and patterns
 Art
 Rhythm
 Visual
 Creativity
 Synthesis

Left side
 Reasoning
 Language
 Writing
 Reading
 Logic
 Mathematics
 Linear
 Analysis

For his split-brain research, Roger Sperry shared
the 1981 Noble Prize in Physiology and Medicine
with David Hubel and Torstein Wiesel.
 if the two hemispheres of the brain are
separated by severing the corpus callosum
the transfer of information between the
hemispheres ceases.
 the coexistence in the same individual of two
functionally different brains can be
demonstrated.

https://coggle.it/diagram/XbSj-afCTQd7nSVk/t/neuroscience-research-contribution-to-cs

https://deepindex.org/

BCI

Logic gates

Microcontrollers

RTOS

Reinforcement Learning

Gestural

Computer Vision

Lederman & Klatzky

channels

Robotics

Developmental robotics

Human-robot interaction

Robotic Voice

Artificial Emotions

Facial Expression

Social Intelligence

Gestures

Personality

Speech Recognition

Evolutionary robotics

Behavior-based robotics

Cybernetics

Cognitive

Educational

Swarm

Soft

Representations

Knowledge Reasoning / Knowledge Representation

Planning

Probabilistic Planning

Domain Independent Planning

Preference-based Planning

Classical Planning

Temporal Planning

Planning Domain Modelling Languages

Ontologies

Automated Reasoning Engines

Theorem Provers

Inference Engines

Expert System

Semantic Nets

Frames Rules

System Architecture

Scene Reconstruction

System methods

Image-understanding Systems

Image Restoration

Recognition

Optical character recognition (OCR)

Optical character recognition

Pose estimation

Content-based image retrieval

Handwriting recognition

2D Code reading

Shape Recognition Technology (SRT)

Facial recognition

Haar features (Viola & Jones)

Motion Analysis

Tracking

Egomotion

Optical flow

Image Generation

GANs

Clustering

Expectation Maximization

k-Medians

Hierarchical Clustering

Grid-Clustering

k-Means

DBSCAN

Agent mining

Text mining

Anomaly/outlier/change detection

Genetic algorithms

Structured data analysis

Intention mining

Big data

Association rule learning

Sequence mining

Time series analysis

Natural language processing

Syntax

Terminology extraction

Lemmatization

Stemming

Parsing

Part-of-speech tagging

Word segmentation

Morphological segmentation

Sentence breaking

Semantics

Lexical semantics

Machine translation

Topic segmentation and recognition

Question answering

Natural language understanding

Word sense disambiguation

Optical character recognition (OCR)

Named entity recognition (NER)

Natural language generation

Recognizing Textual entailment

Sentiment analysis

Relationship extraction

Discourse

Coreference resolution

Discourse analysis

Automatic summarization

Speech

Speech segmentation

Speech recognition

Text-to-speech

Chatbots

Averaged One-Dependence Estimators(ADDE)

Rule System

Repeated Incremental Pruning to Produce Error Reduction(RIPPER)

Cubist

One Rule(OneR)

Zero Rule(ZeroR)

Dimensionality Reduction

Principal Component Regression(PCR)

Projection Pursuit

Principal Component Analysis(PCA)

Linear Discriminant Analysis(LDA)

Partial Least Squares Regression(PLSR)

Quadratic Discriminant Analysis(QDA)

Partial Least Squares Discriminant Analysis

Mixture Discriminant Analysis(MDA)

Sammon Mapping

Flexible Discriminant Analysis(FDA)

Regularized Discriminant Analysis(RDA)

Multidimensional Scaling(MDS)

Game Bot

Autonomous Vehicle

Communities

Papers

People

you're on the right way, go on

yup, keep going, it is just kinda far

Google Brain

Computer System for Machine Learning

Tensorflow
TPU
TPU 2.0
TPU 3.0

Rajat Monga
Principal Scientist
Tensorflow/OpenML library

Paul Barhm
Principal Scientist
TPU/Low level Tensorflow

Patrick Nguyen
Senior Staff Research Scientist

Sanjay Ghemawat
Software Engineer
Data Management
Distributed Systems and Parallel Computing
Machine Intelligence
Natural Language Processing
Software Systems

Martín Abadi
Senior Research Scientist

Zhifeng Chen ✅
Senior Software Engineer
Distributed Systems and Parallel Computing
Machine Intelligence
Networking

Shuyuan Zhang ✅
Senior Software Engineer
Deep Learning, speech to speech translation

Neal Wu ✅
Research Software Engineer
Tensorflow

Youlong Cheng ✅
Sottware Engineer
Street View Infrastructure
FasterRCNN and ResNet TPU model

Frank Chen ✅
Research Software Engineer
Tensorflow

Richard Wei ✅
Software Engineer
Tensorflow

Yanping Huang ✅
Software Engineer
AutoML,Tensorflow performance improvement and distributed multi-GPU ML system

Jue Wang ✅
Tech Lead Google Cloud TPU
flow integration and visulation

Jamin Chen ✅
Senior Software Engineer
Distributed Systems and Parallel Computing
Machine Intelligence

Yuanzhong Xu ✅
Senior Software Engineer
System for Machine Learning,TPU

Bjarke Hammersholt Roune
Staff Software Engineer

Max Galkin
Senior Software Engineer

Xiaobing Liu ✅
Staff Software Engineer
Tensorflow
Distributed System and Parallel Computing
Machine Intelligence
Machine Translation NLP

Mustafa Isir
Staff Software Engineer
Tensorflow

Robert Hundt
Distinguished Engineer
SW Lead for Google TPU and Cloud TPU

Natural Language Understanding/Perception

Sequence to Sequence Learning
with Neural Networks (2014)
A Neural Network for Machine Translation, at Production Scale
Zero-Shot Translation with Google’s Multilingual Neural Machine Translation System
Tacotron 2’s model architecture

Yonghui Wu ✅
Principal Scientist
-Information Retrieval; Ranking; ML
-NLP
-Machine translation

Samy Bengio
Principal Scientist
-Statistical ML

Lukasz Kaiser
Staff Research Scientist

Ciprian Chelba
Research Scientist

Amarnag Subramanya
Senior Research Scientist

Amir Globerson
Research scientist

Nevena Lazic
Research Scientist

Tara Sainath
Senior Staff Research Scientisit

Rohit Prabhavalkar
Senior Research Scientist

Fernando Pereira
Distinguish Scientist in NLP and ML

William Chan ✅
Research Scientist
ML/DL/NLP

Quoc V. Le
Senior SW Engineer NLP

Patrick Nguyen
Senior Staff Research Scientist

Daniel Gillick
Senior Research Scientist

Thang Luong ✅
Research Scientist
Nueral Machine Translation

Machine Learning Algorithm and Techiques

capsules
sparsely-gated mixtures of experts
hypernetworks
new kinds of multi-modal models
-symbolic and non-symbolic learned optimization methods
back-propagate through discrete variables

Geoffrey Hinton
Capsule Neural Network
Co- founder/ principical scientist

Algorithm and theory-
-Machine intellengce
-NLP
-Speech processing

William Chan ✅
Research Scientist
ML/DL/NLP

Noam Shazeer
Research Scientist

Jascha Sohl-dickstein
Staff Research Scientist

Sara Sabour
Research SWE

Ashish Vaswani
Senior Research Scientist

Noam Shazeer
Research Scientist

Lukasz Kaiser: Staff Research Scientist

David Sussillo
Research Scientist

Jakob Uszkoreit
Senior Staff Software Engineer

Oriol Vinyals
Research Scientist

Barret Zoph
Research Scientist

Azalia Mirhoseini
Research Scientist

Nick Frosst
Research SWE

Healthcare and Bioscience
Ophthalmology and digital pathology

Diagnosing Diabetic Eye Disease
Assisting Pathologists in Detecting Cancer
Genomics: DeepVariant Highly Accurate Genomes with DNN.

Greg Corrado
Senior research scientist/Co Founder of Google Brain

Biological Neuroscience
-Artificial Intelligence
-Scalable ML

Navdeep Jaitly
Senior Research Scientis

Derek Wu ✅
Software Engineer
Machine Intelligence/Machine Perception

Yuhui Chen ✅
Software Engineer
xtracting structural information from images/texts

Yi Zhang ✅
Staff Software Engineer, Tech Lead of Health Research
Data injection and ETL
Data Model Infrastruture

Tejas Sundaresan
Software Engineer

David Fiala
Security and Privacy Lead of Health Research

Akosua Busia
Google Brain Research Resident

Arunachalam Narayanaswamy
Senior Software Egnineer

Varun Gulshan
Research Scientist Engineering
Co-founder of the Medical Imaging team

David Coz
Staff Software Engineer
Deep Learning

De Wang ✅
Senior Software Engineer
ML modeling infrastructure and Data Platform
Security and Privacy for ML system

Nan Du ✅
Research Scientist
Deep learning and NLP applied to Healthcare

Michael C. Muelly, M.D:
Research Scientist for ML for healthcare

Kai Lu ✅
Software Engineer

Mark DePristo:
Head of Deep Learning for Genetics and Genomics

Jason Hipp
Research Scientist
Lead Pathologist and Clinical Scientist

Martin Stumpe
Technology Leader/ Engineering Manager in Medical AI

Yang Li Hector Yee ✅
Staff Software Engineer
Deep Learning to Healthcare

Philip Nelson
Director of Engineering

Scott Mayer McKinney
Research Engineer
ML

George Dahl
Research Scientist;
Deep Learning Approach to Linguistic and Perceptual Data,Chemical, Biological, Medical Data

Kun Zhang ✅
Health Research
Recommendation System Machine Learning
Distributed System

Linh Tran
Research Scientist
Machine Learning and Deepl Learning Algorithm for pediatric nervous system

Ting Wang ✅
Senior Software Engineer at Google

Robotics

Deep Reinforcement Learning for Vision-Based Robotic Grasping
Time-Contrastive Networks:
Self-Supervised Learning from Multi-View Observation

Vincent Vanhoucke
Principal Scientist
-Machine Intellegence
-Machine Precption
-Robotics

Alexander Toshev
Research Scientist

Krzysztof Choromanski: Research Scientist

Julian Ibarz:
Techinial Lead

Jasime Hsu ✅
Software Engineer
Machine Inteligence

Caroline Pantofaru
Staff Research Scientist

Vikas Sindhwani
Staff Research Scientist
Mathematical foundations and algorithm design in building robust and scalable learning/ Machine Perception

Kai Kohlhoff
Senior Research Scientsit/ technical lead of Robot Intellegence

Erwin Coumans:
Creator of Bullet Physics: Physics Simulation, Robotics, Deep Reinforcement Learning, Virual Reality

Ted Xiao ✅
Research Engineer
Deep Reinforcement Learning on TIme contrastive Network

Eric Jang ✅
Research Engineer
Machine Intelligence/Machine Perception

Anelia Angelova
Research scientist/ CV, deep learning robotics and perception

Peter Pastor Sampedro
Roboticist
Robotics, Algorithm, Machine Learning, Mobile Manipulation

Sudheendra Vijayanarasimhan
Software Engineer
Video Classification; Action Recognition

Pierre Sermanet:
Research Scientist
deep learning and CV and robotics

Jonathan Shen ✅
Research Engineer
Deep Learning in CV/AR

Yunfei Bai ✅
Technical Lead google robotics
physics-based simulation
robot manipulation

Matthew Kelcey
Senior Researcher DL Reinforcement for robotics

Jie Tan ✅
Software Engineer
Deep Learning /Reinforcement Learning/Robotics/CV/SLAM/ 3d Scanning system/depth map gerneration

Rahul Sukthankar
Senior Rsearch Scientist/Principal Scientist

Navdeep Jaitly
Senior Research Scientist

Tim Blakely
Technical Lead SW Engineer

Karol Hausman
Research Scientist
interactive perception, reinforcement learning and probabilistic state estimation

Montserrat González Arenas
Senior Software Engineer

Marek Fišer
Researcher

Sean Augenstein
Senior Research Scientist

Dumitru Erhan
Senior Research Scientist

Hartmut Neven:
Technical lead manager for Robotics in CV

Ken Caluwaerts: Research Scientist

Mrinal Kalakrishnan
Senior Robotics Scientist

Marek Fišer
Software Engineer
Reinforcement Learning

Aleksandra Faust: Researcher; Lead of motion planning and deep learning task

Corey Lynch
Google Research Resident

PAIR: Big Picture Visualization Group

Facets/ Facets Dive
Distill
Quick, Draw!

Fernanda Viegas:
Senior Research Scientist- big picture group- data visualization
-Distributed systems and parallel computing
-Education innovation
-Human- Computer Interaction and Visualization
-Machine Intelligence

Martin Wattenberg
Research Scientist/Co-founder of big picture group

James Wexler
Senior Software Developer

Ed Chi: ✅
Principal Scientist for Neural Recommender System and Social Computing Research
improvements of recommenders for YouTube, Google Play Store and Google+

Yonghui Wu ✅
Principal Software Engineer
ML and DL for ranking system
automatic speech translation
speech synthesis

Quoc V. Le
Research Scientist

H. Brendan McMahan
Research Scientist

Mike Schuster
Senior Staff Research Scientist

Jess Holbrook
Senior Staff UX Researcher

Josh Lovejoy
Staff User Experience Designer

Mahima Pushkarna
Visual Designer Lead

D. Sculley
Senior Staff Software Engineer

Music and Art Generation
Megenta project
Developing new deep learning and reinforcement learning algorithms for generating songs and other art materials

Douglas Eck
Principal Scientist
-human computer interaction and visualization
-Machine intelligence
-Machine Perception
-NLP

Samy Bengio
Research Scientist Statistical Machine Learning

Thang luong ✅
Research Scientist
NLP
Dialog System/Reading Comprehensive

Curtis Hawthrone
Senior Software researcher on Magenta

Adam Robert
Senior Software Engineer/Megenta team

David Ha
Research Scientist

Erich Elsen
Research Scientist
Tensorflow Development and Reinforcement learning research on Music and Art

Pablo Samuel Castro
Senior Research Software Engineer
Reinforcement Learning and ML applied to music and creativity

Jesse Engel
Research Scientist
Creative Applications of DL/Magenta

https://coggle.it/diagram/XbTHY-fHayl_rmbP/t/google-brain

Schools of thought

Contemporary

Scientific Method

Inductive

Popper (il pue)

People

William Wundt

basic elements of thought
-Sensations:basic elements of perception
-Feelings :basic elements of emotion

https://tinyurl.com/courants-psychocog

Jacques Lacan

Ferdinand de Saussure

Claude Levi-Strauss

Associationism (1880-1920)

People

What ?

Methods

Limits

Analytic Introspection
“Observation of one’s own thought processes”
Required training
High confirmation bias
Unreliable and not objective.

Edward Lee Thorndike

What ?

Methods

Limits

Hermann Ebbinghaus

Behaviourism (1920-1960)

What ?

Methods

Limits

People

Burrhus Skinner

Ivan Pavlov

John Watson

Edward Tolman

Karl Lashley

Constructivism (1923-?)

What ?

Methods

Limits

People

Jean Piaget

Gestalt (1920-1950)

What ?

Methods

Limits

People

Kurt Koffka

Wolfgang Köhler

Max Wertheimer

Kurt Lewin

Cognitivism (1950-?)

What ?

Methods

Limits

People

Noam Chomsky

Georges Miller

Jerome Bruner

Jerry Fodor

Herbert Simon

Allan Newell

Jean Piaget

Claude Shannon

Warren Weaver

Marvin Minsky

John McCarthy

Connexionism (1970-?)

What ?

Methods

Limits

People

Frank Rosenblatt (American psychologist)

Warren McCulloch

Walter Pitts

Donald Hebb

David Rumelhart

David Parker

Embodied Cognition (1990-?)

What ?

Methods

Limits

People

Francisco Varela

Maturana

Evan Thomson

Eleanor Rosch

children learn language through operant conditioning

Children imitate speech they hear

Correct speech is rewarded

language must be determined by inborn biological programming

Children say things that:
They have never heard and can not be imitating
That are incorrect and have not been rewarded for

Rat maze experiment

Evolutionnary psychology

Social Cognition

Types of Learning

Operant Conditioning (Skinner)

Punishment

Negative Reinforcement

Positive Reinforcement

Reinforcement

Positive Reinforcement

Negative Reinforcement

Classical Conditioning (Pavlov)

Cognitive Psychology

Developmental & Learning Psychology

Psychopathology

Psychanalisis

Sociolinguistics

Ethnolinguistics

Cooing stage

Developmental linguistics

Intention seeking

Pattern Finding

Views/Approaches

Early Grammarians

Structuralism

Generativism

Functionalism

Cognitive linguistics

Semantics

Generative

Semio

Semiotics

Semiology

Justine Cassell

Translation / NLP / Computational Linguistics

Clinical linguistics

Evolutionary linguistics

Pāṇini Sanskrit morphology

cuneiform clay tablets

Philology

study of language in oral and written historical sources; it is the intersection of textual criticism, literary criticism, history, and linguistics.

Grece and the Stoics

Sibāwayhi arabic linguistics

Irish Sanas Cormaic 'Cormac's Glossary' encyclopedic dictionary

'De vulgari eloquentia' Dante

Ferdinand de Saussure

diachronic

synchronic

syntagme

paradygme

The Prague school

Roman Jakobson

6 language functions

Leonard Bloomfield

Louis Trolle Hjelmslev

Noam Chomsky

George Lakoff

Linguistic Wars

Linguistic turn

continental post structuralism VS logic and analytic philosophy of language

Augmented Reality

Language aquisition

Roland Barthes

Charles Sanders Peirce

Umberto Eco

signifiant / signifié

Pragmatisme

abduction

against cartheianism

triad

representamen

interpretant

object

novels

Ethnology

Ethnography

ML Optimizers

gradient descent

SGD (calculated in every sample of the ds per epoch)

MBGD (same than SGD but with mini batchs)

NAG (reduce error by using futur steps)

Momentum (throwing ball downhill, go faster if in right direction)

Adaptive learning rate

Adagrad (adapts learning rate to the parameters based on previous gradients, but has learning rate shrinking pb)

RMSProp (decaying average of all past square gradients)

Adam (RMSProp + bias correction + momentum)

Adadelta (decaying average of all past square gradients)

Nadam (Adam + NAG)

BGD (calculated on entire ds)

Natural Sciences

Design

Education and training

psychology

anthropology

physiology

industrial design

typography

architechture

CAD

lazer cutting

additive manufacturing

material engineering

electrical engineering

mechanical engineering

WIMP Windows Icon Menu Pointing

Multimodal

gestural channel

semiotic

epistemic

Fitz Law

ergotic

proprioception

touch

kinestesia

Modes

other types of interactions

"beating Fitz Law"

Space scale diagrams

bubble cursor

zoomable UI

multiscale pointing

Goal passing

Steering Law

Conceptual Models

Affordances / Signifiers
(JJ Gibson/ Don Norman)

Metaphors

3 rules (Don Norman)

3 dimentions of an interface

Reuse (interaction/visualisation)

Reification (objects)

Polymorphysm (commands)

visibility

mapping

feedback

Google Deepmind

Memory

Attention

Conciousness

Arousal ("vigilance")

Decision Making

Natural Language

Learning

Motor coordination

Perception

Planning

Problem Solving

Thoughts