1.3 Six Elements of ML
A defining framework for understanding concepts in the course
Recap: Machine Learning
What we saw in the previous chapter?
(c) One Fourth Labs
Repeat the last slide of the previous chapter
A jargon cloud
How do you make sense of all the jargon?
(c) One Fourth Labs
Make an actual cloud of all keywords that we will see through the course (list down all the keywords from the table of contents on my course homepage
From jargons to jars
What are the six jars of Machine Lerarning
(c) One Fourth Labs
Show six empty or shaded jars
* I want images which look like this but this is an expensive image and not available for free
Data data everywhere
What is the fuel of Machine Learning?
(c) One Fourth Labs
Data data everywhere
How do you feed data to machines ?
(c) One Fourth Labs
We encode all data into numbers - typically high dimension
For instance, in this course you will learn to embed image and text data as large vectors
Data entries are related - eg. given a MRI scan whether there is a tumour or not
Include a table that shows two/three MRI scans in first col, shows large vectors in second column, 1/0 for last column of whether there is tumour or not
Include a table that shows two/three reviews in first col, shows large vectors in second column, 1/0 for last column for whether review is positive or negative
Title the columns as x and y
tumor/no tumor
Now show a matrix of numbers here
the last column is tumor/no tumor
an arrow here
Show the ML system from your Slide 10 of Expert Systems
an arrow here
Data curation
Where do I get the data from?
(c) One Fourth Labs
I am lucky
I am rich
I am smart
+ मुंबई
= मुंबई
In this course
Data data everywhere
What is the fuel of Machine Learning?
(c) One Fourth Labs
Show data jars
Tasks
What do you do with this data?
(c) One Fourth Labs
We want to show that we can do different tasks with same data by making different input output papers. Show the following input output pairs one-by one. Feel free to redefine the input output pairs given below suitably
1. From product description to structured specs
2. From specs + reviews to writing FAQs
3. From specs + reviews + FAQs to question answering
4. From specs + reviews + personal data to recommendations
Amazon product data with description, reviews, product specs
Input
Output
Tasks
What do you do with this data?
(c) One Fourth Labs
We want to show that we can do different tasks with same data by making different input output papers. Show the following input output pairs one-by one. Feel free to redefine the input output pairs given below suitably
1. From photos identify people, places, activities
2. From posts + personal data recommend posts
3. From video detect profanity, etc.
Facebook profiles and photos
Input
Output
Tasks
What do you do with this data?
(c) One Fourth Labs
Different types of tasks:
1. Supervised
- Classification - text or no text
- Regression - fitting bounding boxes (more later)
2. Unsupervised
- Clustering - clustering news articles by similarity
- Generation - deep art, deep poetry
Most of the realworld ML tasks (90%) are supervised. This course will exclusively focus on this class of problems. Except for easter eggs.
In supervised ML it is about finding y given x
Supervised
1. Show data matrix with x and y
2. images with and without signboards
Here show a SVM like line separator with signboard images on one side and no-signboard on other
Classification
Tasks
What do you do with this data?
(c) One Fourth Labs
Different types of tasks:
1. Supervised
- Classification - text or no text
- Regression - fitting bounding boxes (more later)
2. Unsupervised
- Clustering - clustering news articles by similarity
- Generation - deep art, deep poetry
Most of the realworld ML tasks (90%) are supervised. This course will exclusively focus on this class of problems. Except for easter eggs.
In supervised ML it is about finding y given x
Supervised
1. Show data matrix with x and left_x, left_y, width, height
2. images with signboards and bounding boxes
Image
Regression
Output
lx, lr, w, h values
Now show bounding box in the images
Tasks
What do you do with this data?
(c) One Fourth Labs
Different types of tasks:
1. Supervised
- Classification - text or no text
- Regression - fitting bounding boxes (more later)
2. Unsupervised
- Clustering - clustering news articles by similarity
- Generation - deep art, deep poetry
Most of the realworld ML tasks (90%) are supervised. This course will exclusively focus on this class of problems. Except for easter eggs.
In supervised ML it is about finding y given x
Unsupervised
1. Show data matrix with only x
2. images with and without signboards
Here show 3 to 4 clusters such that yellowfins signboard in one cluster, blue in another and so on
Clustering
Tasks
What do you do with this data?
(c) One Fourth Labs
Different types of tasks:
1. Supervised
- Classification - text or no text
- Regression - fitting bounding boxes (more later)
2. Unsupervised
- Clustering - clustering news articles by similarity
- Generation - deep art, deep poetry
Most of the realworld ML tasks (90%) are supervised. This course will exclusively focus on this class of problems. Except for easter eggs.
In supervised ML it is about finding y given x
Unsupervised
1. Show data matrix with only x
Show picasso style images
Show output of deep art or of the painting which recently got sold for x million dollars
Generation
Tasks
What do you do with this data?
(c) One Fourth Labs
Different types of tasks:
1. Supervised
- Classification - text or no text
- Regression - fitting bounding boxes (more later)
2. Unsupervised
- Clustering - clustering news articles by similarity
- Generation - deep art, deep poetry
Most of the realworld ML tasks (90%) are supervised. This course will exclusively focus on this class of problems. Except for easter eggs.
In supervised ML it is about finding y given x
Unsupervised
Show many trump tweets
Show output of Deep Trump
Generation
Tasks
What do you do with this data?
(c) One Fourth Labs
\( `` \)
Photo of Andrew Ng
Supervised Learning has created 99% of economic value in AI
In this course
Classification
Regression
RHS from classification slide
RHS from regression slide
Tasks
What do you do with this data?
(c) One Fourth Labs
Show data, tasks jars
What is the mathematical formulation of a task?
(c) One Fourth Labs
\( x \)
\( y \)
bat
car
dog
cat
Models
\( \left[\begin{array}{lcr} 0.2, 0.1, 0.7, ......0.8 \end{array} \right]\)
\( \left[\begin{array}{lcr} 0, 0, 1,0, 0 \end{array} \right]\)
\( y = f(x) \) [true relation, unknown]
\( \hat{y} = \hat{f}(x) \) [our approximation]
ship
\( \left[\begin{array}{lcr} 0, 0, 1,0 \end{array} \right]\)
Models
What are the choices for \( \hat{f} \) ?
(c) One Fourth Labs
- Show some points sampled from this function
- Say that there is some complex relation between x
- Naively I assumed that its y=mx + c
- no matter how I adjust m and c I can't make f and \( \hat{f} \) equal (apparently you can make a video of this in python itself)
- Let's try another function which is a polynomial
- another function...better...better better
\( y = mx + c \)
\( y = ax^2 + bx + c \)
\( y = \sigma(wx + b) \)
\( y = Deep\_NN(x) \)
\( \hat{y} = \hat{f}(x) \) [our approximation]
\( \left [\begin{array}{lcr} 0.2\\ 0.1\\ 0.7\\ ....\\0.8 \end{array} \right]\)
\( \left [\begin{array}{lcr} 2.2\\ 3.1\\ 0.7\\ ....\\4.8 \end{array} \right]\)
\( x \)
\( y \)
\( y = ax^3 + bx^2 + cx + d \)
\( y = ax^4 + bx^3 + cx + d \)
I want to say that the true data was drawn from this function
These should be the points drawn from the function shown
Use Y_hat everywhere
In this course
\( y = Deep\_CNN(x) \) ...
\( y = RNN(x) \) ...
Models
Why not just use a complex model always ?
(c) One Fourth Labs
\( \left [\begin{array}{lcr} 0.2\\ 0.1\\ 0.7\\ ....\\0.8 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.4\\ 0.2\\ 1.4\\ ....\\1.6 \end{array} \right]\)
\( x \)
\( y \)
This will be replaced by a simple line
We will show animation how it will be easy to fit a line but difficult to fit 100 degree polynomial
\( y = mx + c \) [true function, simple]
\(y = ax^{100} + bx^{99} + ... + c \) [our approximation, very complex]
Later in this course
Bias-Variance Tradeoff
Overfitting
Regularization
Models
What are the choices for \( \hat{f} \) ?
(c) One Fourth Labs
Add model jar
Loss Function
How do we know which model is better ?
\( \left [\begin{array}{lcr} 0.00\\ 0.10\\ 0.20\\ ....\\6.40 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.24\\ 0.08\\ 0.12\\ ....\\0.36 \end{array} \right]\)
\( x \)
\( y \)
?
\( \hat{f_1}(x) \)
\( \left [\begin{array}{lcr} 0.25\\ 0.09\\ 0.11\\ ....\\0.36 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.32\\ 0.30\\ 0.31\\ ....\\0.22 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.08\\ 0.20\\ 0.14\\ ....\\0.15 \end{array} \right]\)
\( \hat{f_1}(x) = 1.79x^{25} - 4.54 x^{24} + ... - 1.48x + 2.48 \)
\( \hat{f_2}(x) = 2.27x^{25} + 9.89x^{24} + ... + 2.79x + 3.22 \)
\( \hat{f_3}(x) = 3.78x^{25} + 1.57x^{24} + ... + 1.01x + 8.68 \)
\( \begin{array}{lcr} 1\\ 2\\ 3\\ ....\\n \end{array} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
\( \hat{f_2}(x) \)
\( \hat{f_3}(x) \)
\( \mathscr{L}_2 = \sum_{i=1}^{n} (y_i - \hat{f}_2(x_i))^2 \)
\( \mathscr{L}_3 = \sum_{i=1}^{n} (y_i - \hat{f}_3(x_i))^2 \)
True Function
\( \hat{f_1}(x) \)
\( \hat{f_2}(x) \)
\( \hat{f_3}(x) \)
why not use numbers ?
whose function is better?
?
Loss Function
How do we know which model is better ?
\( \left [\begin{array}{lcr} 0.00\\ 0.10\\ 0.20\\ ....\\6.40 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.24\\ 0.08\\ 0.12\\ ....\\0.36 \end{array} \right]\)
\( x \)
\( y \)
?
\( \hat{f_1}(x) \)
\( \left [\begin{array}{lcr} 0.25\\ 0.09\\ 0.11\\ ....\\0.36 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.32\\ 0.30\\ 0.31\\ ....\\0.22 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.08\\ 0.20\\ 0.14\\ ....\\0.15 \end{array} \right]\)
\( \hat{f_1}(x) = 1.79x^{25} - 4.54 x^{24} + ... - 1.48x + 2.48 \)
\( \hat{f_2}(x) = 2.27x^{25} + 9.89x^{24} + ... + 2.79x + 3.22 \)
\( \hat{f_3}(x) = 3.78x^{25} + 1.57x^{24} + ... + 1.01x + 8.68 \)
\( \begin{array}{lcr} 1\\ 2\\ 3\\ ....\\n \end{array} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
\( \hat{f_2}(x) \)
\( \hat{f_3}(x) \)
\( \mathscr{L}_2 = \sum_{i=1}^{n} (y_i - \hat{f}_2(x_i))^2 \)
\( \mathscr{L}_3 = \sum_{i=1}^{n} (y_i - \hat{f}_3(x_i))^2 \)
True Function
\( \hat{f_1}(x) \)
\( \hat{f_2}(x) \)
\( \hat{f_3}(x) \)
why not use numbers ?
whose function is better?
?
Loss Function
How do we know which model is better ?
(c) One Fourth Labs
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 = ? \)
\( \mathscr{L}_2 = \sum_{i=1}^{n} (y_i - \hat{f}_2(x_i))^2 = 2.02\)
\( \mathscr{L}_3 = \sum_{i=1}^{n} (y_i - \hat{f}_3(x_i))^2 = 2.34 \)
In this course
Square Error Loss
Cross Entropy Loss
KL divergence
\( \left [\begin{array}{lcr} 0.00\\ 0.10\\ 0.20\\ ....\\6.40 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.24\\ 0.08\\ 0.12\\ ....\\0.36 \end{array} \right]\)
\( x \)
\( y \)
\( \hat{f_1}(x) \)
\( \left [\begin{array}{lcr} 0.25\\ 0.09\\ 0.11\\ ....\\0.36 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.32\\ 0.30\\ 0.31\\ ....\\0.22 \end{array} \right]\)
\( \left [\begin{array}{lcr} 0.08\\ 0.20\\ 0.14\\ ....\\0.15 \end{array} \right]\)
\( \begin{array}{lcr} 1\\ 2\\ 3\\ ....\\n \end{array} \)
\( \hat{f_2}(x) \)
\( \hat{f_3}(x) \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
\( = (0.24-0.25)^2 + (0.08-0.09)^2 + \newline (0.12-0.11)^2 + ... + (0.36-0.36)^2 \)
\( = 1.38 \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 = 1.38\)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 = ? \)
Loss Function
What does a loss function look like ?
(c) One Fourth Labs
Data
Task
Loss
Model
Learning Algorithm
How do we identify parameters of the model?
(c) One Fourth Labs
Animation:
- first the data matrix appears
- then the model equation with a,b,c as parameters appears
- then the friend appears
- then the loss function appears
- now the red cross appears and then the friend disappears
- now the box for the learning algorithm appears
- now the logo for search appears
- now an animation where the values of a,b,c are adjusted till the loss reaches some low value
- data, model and loss function feed int othe pink box
Show a a matrix with 3 inputs: budget* (0 to 1), box office collection* (0 to 1), action scene time+ (0 to 1)
* the unit is 100 crores so 0.1 means 1 crore (mention this in the head row of the table
+ the unit here is 100 minutes
The output is the imdb rating
Show x and y above the header
\( \hat{f_1}(x) = 3.5x_1^2 + 2.5x_2^{3} + 1.2x_3^{2} \)
cartoon images of 3 friends
\( \hat{f_1}(x) = ax_1^2 + bx_2^{3} + cx_3^{2} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
Show a gear box inside this to indicate learning algorithm
Show adjustable scales for a,b,c and create a python video here if you adjust the scale the loss function value changes and you hit some value for which the error is zero. You can actually cheat by creating the y value using some values of a,b,c so that you can then get 0 error for these values of error
logo for search
Learning Algorithm
How do you formulate this mathematically ?
(c) One Fourth Labs
Animation:
- data, model, pink box, loss function, logo for search and animation for a,b,c appears as it if from previous slide
- now the cross and the message appears
- now the text in red appears (but make it black)
Show a a matrix with 3 inputs: budget* (0 to 1), box office collection* (0 to 1), action scene time+ (0 to 1)
* the unit is 100 crores so 0.1 means 1 crore (mention this in the head row of the table
+ the unit here is 100 minutes
The output is the imdb rating
Show x and y above the header
\( \hat{f_1}(x) = ax_1^2 + bx_2^{3} + cx_3^{2} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
Show a gear box inside this to indicate learning algorithm
Show adjustable scales for a,b,c and create a python video here if you adjust the scale the loss function value changes and you hit some value for which the error is zero. You can actually cheat by creating the y value using some values of a,b,c so that you can then get 0 error for these values of error
logo for search
In practice, brute force search is infeasible
Find \(a, b, c \) such that
is minimized
Learning Algorithm
How do you formulate this mathematically ?
(c) One Fourth Labs
Animation:
Only the green part gets added on this slide (but show it in black except for the tick mark)
Show a a matrix with 3 inputs: budget* (0 to 1), box office collection* (0 to 1), action scene time+ (0 to 1)
* the unit is 100 crores so 0.1 means 1 crore (mention this in the head row of the table
+ the unit here is 100 minutes
The output is the imdb rating
Show x and y above the header
\( \hat{f_1}(x) = ax_1^2 + bx_2^{3} + cx_3^{2} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
Show a gear box inside this to indicate learning algorithm
Show adjustable scales for a,b,c and create a python video here if you adjust the scale the loss function value changes and you hit some value for which the error is zero. You can actually cheat by creating the y value using some values of a,b,c so that you can then get 0 error for these values of error
logo for search
Many optimization solvers are available
\(min_{a,b,c}\)
Learning Algorithm
How do you formulate this mathematically ?
(c) One Fourth Labs
Animation:
Only the green part gets added on this slide (but show it in black except for the tick mark)
Show a a matrix with 3 inputs: budget* (0 to 1), box office collection* (0 to 1), action scene time+ (0 to 1)
* the unit is 100 crores so 0.1 means 1 crore (mention this in the head row of the table
+ the unit here is 100 minutes
The output is the imdb rating
Show x and y above the header
\( \hat{f_1}(x) = ax_1^2 + bx_2^{3} + cx_3^{2} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
Many optimization solvers are available
\(min_{a,b,c}\)
In this course
Gradient Descent ++
Adagrad
RMSProp
Adam
Learning Algorithm
How do we identify parameters of the model?
(c) One Fourth Labs
Add jar for Learning Algorithm
Evaluation
How do we compute a score for our ML model?
(c) One Fourth Labs
Show a matrix for x and y (consider cat, dog, giraffe, lion, tiger, etc. prediction)
Now add a columns y for model predictions
Now show ticks and crosses and show we can compute accuracy (show formula)
End by saying that there are other metrics such as precision, recall, etc.
Standard evaluation (example ImageNet)
Evaluation
What are some other evaluation metrics ?
(c) One Fourth Labs
Show a matrix for x and y (consider cat, dog, giraffe, lion, tiger, etc. prediction)
Now add a set of columns y for model predictions which is a ranked list
Now show ticks and crosses for top-1, top-3
In this course
Accuracy
Precision/Recall
Top-k accuracy
Evaluation
How is this different from loss function ?
(c) One Fourth Labs
#( )
Evaluation
Brake
/Go
__________
#( )
Loss function
\( maximize \)
#( )
____________________
#( ) + #(___)
Evaluation
Should we learn and test on the same data?
(c) One Fourth Labs
Show some training data from before, preferably something containing image classification
\( \hat{f_1}(x) = ax_1^2 + bx_2^{3} + cx_3^{2} \)
\( \mathscr{L}_1 = \sum_{i=1}^{n} (y_i - \hat{f}_1(x_i))^2 \)
Show a gear box inside this to indicate learning algorithm
\(min_{a,b,c}\)
Now show test data
Now show formula for accuracy here
Animation:
Only the green part will com on animation, the rest of it will be shown at the beginning itself
Evaluation
How is this different from loss function ?
(c) One Fourth Labs
Add jar for evaluation
Putting it all together
How does all the jargon fit into these jars?
(c) One Fourth Labs
Show six jars and a foundation (3 blocks for Lin. Alg., Prob., Calculus) on which these jars are placed
Here's the hard part now: Remember the word cloud at the beginning. Now you need to organize them into these 6 jars and foundation:-) It's ok if you miss out a few things. Juts put them on the side. I will see where to fit them. Show the word cloud in the background as faded
Data, democratisation, devices
Why ML is very successful?
(c) One Fourth Labs
Show the same diagram as previous slide
1) On top of Evaluation show "standardised", and show logos of ImageNet, Pascal VOC, WMT
2) On top of Learning Algorithms and loss functions show "improvised"
3) On top of models show "democratised"
4) Now on top of 1st jar write "Abundance"
Typical ML effort
How to distribute your work through the six jars?
(c) One Fourth Labs
Show the same diagram as previous slide
1) Now put this box to cover the last 4 jars
2) Now on top of 1st 2 jars say "Your job"
Connecting to the Capstone
How to distribute your work through the six jars?
(c) One Fourth Labs
You can show the six jars from before and remove the foundation stones to save space. the jars can also be small now.
- The rest of the slide should be as animated below
Mumbai
/
/
मुंबई \( \rightarrow \) Mumbai
\( \sum_{i=1}^{n} (y_i - \hat{f}(x_i))^2 \)
\( -\sum_{i=1}^{n} \log \hat{f}(x_i) \)
Accuracy
Precision/Recall
Top-k accuracy
Assignment
How do you apply the six jars to a problem that you have encountered?
(c) One Fourth Labs
Explain the problem
Give link to the quiz
1. Formulate 3 problems from data.gov.in
2. In the dataturks labelled data, define tasks that you can perform and collect 10 data points for each
// Binary classification of whether there is text
// Detect text with bounding box - is accuracy easy to define here?
suman's copy 1.3 Loss done
By suman banerjee
