Serra Hunter @ UPC

 

 

Ahcène Boubekki

 

 

UCPH, Denmark

Short Resume

     Short Resume

Name:

Ahcène Boubekki

Background:

Mathematics
(complex geometry, algebraic geometry...)

Currently:

Post-Doc @ PTB/TU Berlin

PhD:

Analysis of User Behavior

Previously:

Post-Doc @ UCPH/Pioneer Centre for AI

Researcher @ TU Munich

Researcher/PhD @ DIPF/TU Darmstadt (~Frankfurt)

Researcher/PhD @ Leuphana (~Hamburg)

Post-Doc @ UiT Arctic University of Norway

     Short Resume

Main Research Topics

1. Representation Learning

2. Explanability/Interpretability

1. Representation Learning

2. Explanability/Interpretability

Self-explainable Models

Model Inspection

Learning Representations

Representing for Learning

Previous Research

PhD Thesis

     The mBook Project

Objectives

Evaluate the use of an electronic textbook for history in middle school

Bring new methods to Educational Science

The mBook

     The mBook Project

Some statistics

From January 31st to July 11th 2017

     2,197 sessions
     400 users
     195 pupils (537 sessions)

     The mBook Project

Classical Approaches

Bi-variate factor analysis

Markov Chains

Contributions

     Contributions

Same Data

 

Different Models

 

Different Representations

Content Analysis

Periodic Behaviors

Scrolling Behaviors

Online Behaviors

Is there a correlation between content and motivation?

Do the pupils use the mBook the same way over the week?

Are scrolling patterns correlated with competencies?

What is the influence of the teacher?

Content Analysis

Model: Archetypal Analysis

Repr.: Frequency Vectors

How to use AA instead of factor analysis?

Periodic Behaviors

Model: Mixture of Markov Chains

Repr.: Discrete Sequences

Can we make MMCs temporally aware?

Scrolling Behaviors

Model: Infinite/Bayesian MMC

Repr.: Discrete Sequences

How to do data-driven the model selection?

Online Behaviors

Model: Infinite/Bayesian k-means

Repr.: Spatio-temporal Timeseries

Can we study sessions as spatio-temporal trajectories?

     Trajectories and Online Behaviors

Is it relevant to study sessions as
spatio-temporal trajectories?

* Well-defined ~ invariant to subdivision
                            distinguishes prefixes

None satisfies all the properties

Let's build one!

Sequence of pages

↓

Path in the page graph

 

 

 

 

+

Timestamps

 

 

 

 

 

 

+

Metric on the page graph

 

 

 

 

 

 

 

↓

Spatio-temporal

trajectory

Which trajectory measure ?

Construction

Our Measure:

\Delta_{\mathrm{KL}}( X,Y : \lambda) = (1-\lambda)\log \frac{T_Y}{T_X} + \lambda \frac{1}{T_X} \int_{0}^{T_X} d \big( X(t), Y(t) \big) \partial t

    Digression about

\Delta_{\mathrm{KL}}
\Delta_{\mathrm{KL}}

 with shift

\Delta_{\mathrm{KL}}

without shift

K.V. Olesen, et al. "A Contextually Supported Abnormality Detector for Maritime trajectories." Journal of Marine Science and Engineering  (2023)

Trajectories in same cluster tend to start in the same location

 

What if we allow a temporal shift ?

     Trajectories and Online Behaviors

Positive Correlation ↔ High avg.     ≈ More Freedom

\psi

Negative Correlation ↔ Low avg.    ≈ Less Freedom

\psi

Statistically Significant

Activity Indicator

\psi :=

average distance between one pupil
and her classmates

\mathrm{PPM} :=

number of page per minute

\mathrm{EPM} :=

number of event per minute

Pupils perform better if they follow the teacher's style/instructions

Teacher A

Teacher B

\text{avg.}\psi

Teacher B

Teacher A

     Deep Clustering as a Unifying Method

Haven't we be doing the same thing?

Neural
Networks

Centroids

Neural Networks ?

Model Objects Dis/similarity Clustering
AA Vectors Euclidean + kernel Centroids
MMC Matrices Probability Centroids
Trajec. Time series Temporal simil. Centroids

Embedding

From Clustering to
Deep Clustering

A Short Lesson

     Affinity-based Clustering

We cannot make everyone happy !

Problem:

Group super heroes

Objective:

Everyone is happy

Minimize unhappiness

     Affinity-based Clustering

0.6

0.4

0.3

0.9

0.1

0.6

0.7

0.8

0.9

0.1

0.4

0.2

0.1

0.9

0.1

0.2

0.3

0.7

0.5

0.8

0.9

0.5

0.4

0.3

Memory
expensive

How many
groups?

Where should we cut the graph?

✂

     Affinity-based Clustering

Group together those that are clearly similar

Strategy:


and treat the rest as noise.

DBSCAN

Ester, Martin, et al. "A density-based algorithm for discovering clusters in large spatial databases with noise." kdd. Vol. 96. No. 34. 1996.

     Affinity-based Clustering

Group together those that are clearly similar

Strategy:

 

Merge if one member is similar enough to one other member.

 

 

Until enough is not satisfied anymore.

 

 

Until 3 clusters are formed.

Agglomerative Clustering

 

Single-linkage


Different merging strategy,
different linkage

 

still queries the Affinity matrix
(N×N)

     Affinity-based Clustering

Remarks:

Which similarity measure?

 

Euclidean distance is easy

 

 

Not always cluster vectors

Cost of the
affinity matrix

 

Compute over mini-batches

 

 

Might repeat computations

Objects don't move!

 

The decision borders move

 

 

Let's make the objects move!

     Affinity-based Clustering

Euclidean distance is easy

Compute over mini-batches

Let's make the objects move!

Euclidean distance is easy

Compute over mini-batches

Let's make the objects move!

color

shape

What are we actually doing?

 

We learn a similarity measure

 

We learn a Kernel!

Feature map

Euclidean

Unknown

 

How do we guide the learning?

     Affinity-based Deep Clustering

Let                                         be a dataset that we want to cluster using a feature map                       .

\mathcal{X}\!=\!\{x_1\ldots x_N\}\!\subset\!\mathbb{R}^d

We want that in the embedding space:

    - similar objects are close to each other,

    - dissimilar ones are far from each other.

f\!:\!\mathbb{R}^d \!\longrightarrow \!\mathbb{R}^p

For each datapoint    , we have a set of positive examples    
and of negative ones      .

x_i
x_i^{+}
x_i^-
\mathcal{J} = \sum_{i=1}^N \sum_{j \in x_i^+} ||x_i -x_j||^2 - \sum_{l \in x_i^-} ||x_i -x_l||^2

Triplet Loss

How do we get these sets?

Euclidean norm not good in practice

\mathcal{J} = \sum_{i=1}^N -\log\Big( \dfrac{ \sum_{j \in x_i^+} \exp \cos( x_i, x_j) / \tau }{ \sum_{l \in x_i^-} \exp \cos( x_i, x_l) / \tau } \Big)

InfoNCE

     Contrastive Learning

Augmentations are pulled closer

Other instances are pushed away

     Centroid-based Clustering

Choose three representatives.

Strategy:

 

Group by similarity.

 

 

Update the representatives.

 

 

 

Continue until convergence.

k-Medoids

Can we learn k-means with a neural network?

If the representatives are not necessarily instances

 

k-means

Research Plan

Mechanistic Interpretability

     Motivation

What did you see?

Where did you look?

Where were

the eyes?

How many eyes

were there?

How to see what
a model sees?

     How to see what a model sees?

RISE

XRAI

GradCAM

LRP

IG

Dingo or Lion

These do not answer directly
what does the model see?

Deep Dream?

What makes it more a tiger than a tiger?
Too slow, impossible to train, not really useful.

What is important for the prediction?
Inconsistent, difficult to read, objective unclear.

Saliency Maps?

How is the neighborhood in the embedding?
Inspection of the embedding, "biaised" justification..

Prototypes/Concepts?

Counterfactual?

What should I change to change class?
Tricky to compute, but nice!

     How to see what a model sees?

Standard Image Classifier

\rbrace

Encoder

convolutions, pooling, non-linearity, skip-connections, attention, etc.

\rbrace

Classifier

Single linear layer... eventually a softmax

clustering

k=10

k=5

     How to see what a model sees?

Standard Image Classifier

\rbrace

Encoder

convolutions, pooling, non-linearity, skip-connections, attention, etc.

\rbrace

Classifier

Single linear layer... eventually a softmax

clustering

k=10

k=5

     How to see what a model sees?

Seems like a déjà-vu?

One object at a time

Limited to 3 directions

For K=3,
PCA and k-means are similar

K-means provides
some hierarchy!

What can we "explain"?

     What can we explain?

Explain explanations

IG

LRP

GradCAM

RISE

XRAI

color gradient ~ rank

     What can we explain?

Connect Concepts and Semantics

     What can we explain?

ECG Explanation

     What can we explain?

Annotation Masks

Model train for binary classif.

 

All lesions recovered

 

 

Can we use NAVE for medical annotations?

 

 

You need well performing model!

     What can we explain?

Inspect Shortcuts Saturation

What else
can we do?

     What else can we do?

Unlearning?

>>> Test set
>>> IMPUTATION: SWAP

>>> Accuracies
                 All    LBL==0  LBL==1
Distribution    176     26.7    73.3
--------------------------------------
Nothing      :  26.7    100.0   0.0
Watermark    :  100.0   100.0   100.0
Wmk and Imput:  26.7    100.0   0.0
--------------------------------------

     What else can we do?

In distribution counterfactuals

Future Research

     Future Research

Theory

  • Health data: ECG and mammography on going.
  • Transfer to generative models, RL, LLM.

Applications

  • Connect NAVE and feature importance.
  • Do causality.
  • Counterfactual.
  • Unlearning.

How to use NAVE for Fairness and Trustworthiness?

Serra Hunter @ UPC

 

 

Ahcène Boubekki

 

 

UCPH, Denmark