ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
2
Spring Semester 2024: Feedback
3
DateLocationlesson #ResponsibleS24 Topics We coveredWe need to follow up:We liked thatWe could improveThe homework is:
4
Person who is in charge of provide the feedbackThese are just examples or suggestions, please feel free to use this space as much as you wantTopics were covered...
Pace was just right...
Most of the topics were covered except...		These topics maybe need to be reviewed in the first minutes of the next classes...
Learners did a lot of questions regarding this *topic*....		Learners looked engaged...
Learners were participative!
All learners completed the task successfully!
Exercises were right for the topic		The task is too complex....
We need to lower the pace...
This topic is better if we explain it onsite...
Maybe we can use this *tool* next time		Teacher Owner assign the next homework:
5
6
Wed, Mar 13📍On-site1Nathan3We covered the course outline for the first semester. We completed some ice breakers. We instructed the students what the need to install to be ready for next weeknoneSome students looked a little bored midway through but everyone kept focused and it went wellnonecheck that the basic package are installed and ready to go
7
Mon, Mar 18💻 Online2Nathan4History of AI, Python warm up notebook, Bias, AI TechniquesWould be nice to have follow up steps for the Python warm-upWe've covered theoretical material and have a basic understanding of level and speed of studentsChecking in with students who raise thier hand to ask questionsFinish exercises and refresh Python knowledge
8
Wed, Mar 20📍On-site3Nathan3History of AI, ML fundamentals, course overview , what will be taught in the futurestudents really stayed engaged none
9
Mon, Mar 25💻 Online4Olena3Recap on the tooling (really quick)
Managing data
Linear regression
Polynomial regression
Slides can be found S24 ML&AI: 4. ML Regression (Part 1))		- Nothing (invited attendees to bring any outstanding questions to slack)- A lot of good questions from the attendees, though significantly more questions from male participants comparing to women participants
- Based on the raised questions - seems that many participants followed the practical excercises
- Good idea to ask students to ask out loud if they have a question, instead of writing in the chat!		- I(Olena)'d increase some of the fonts on the slides and even make more slides with less info per slide
- Excercise # 4 from S24 ML&AI: 4. ML Regression (Part 1)
10
Wed, Mar 27📍On-site5Olena3Overfitting and Underfitting
Regularization
Cross Validation
Hyperparameter Tuning
Slides can be found in S24 ML&AI: 5. ML Regression (Part 2))
nothing- Active participants
- warm atmosphere
- great presentation by the teacher		- for in-person classes we can encourage works in teams
- request to get class materials a bit in advance, so that students can prepare better at home		 no homework, but recommendation to check Kaggle datasets and competitions
11
Mon, Apr 1Public Holiday
12
Wed, Apr 3📍On-site6HasnainRecap of ML Regression, Boston Housing Dataset, Feature Engineering, Decision Trees and Random Forests
- Fitting a few other regression models
on boston housing (Random Forest, XGB)
Feature Engineering Topics:
- Normalizing and scaling data
- Imputation
- Handling outliers
- Binning
- Log Transform
- Encoding (One hot)
- Grouping
- Splitting
- Date processing
- PCA
Evaluation:
- GridSearchCV
Nothing- Students raised their hands a lot, good participation- A lot of students didn't show up (maybe the rainy weather got them down)
- Not everyone seems super engaged, especially people in the back
- I (Hasnain) found it hard to engage with the students, maybe that's because I'm distracted with the feedback submitter duties.
13
Mon, Apr 8💻 Online7HasnainClassification, Confusion Matrix- Introducing ML Classification through some basic examples
- Binary and MultiClass
- What are TP, FP, TN, FN and accuracy
- Constructing a confusion matrix
- ROC-AUC
- Dealing with data imbalance		Nothing- The Gunes classifier was fun and engagingNo homework, but recommendation to go through the notebooks
14
Wed, Apr 10📍On-site8- Class material on google classroom
- Classification recap
- Algorithms
1. Logistic Regression
2. Decision Trees
3. Random Forests
4. Support Vector Machine		- usage of white board to explain with examples.
- students asking questions
- well structured class with concepts clearly explained by Francisco
- class material shared in advance as per previous student feedback 		-Student attendance was low. We could use student feedback to understand why. Homework on Google classroom. Advised to do based on in-class exercise.
15
Mon, Apr 15💻 Online9AzadClustering-ML Clustering
-DimRed ML Clustering
-DimRed Clustering Techniques
1.kmeans,
2.hierarchical clustering,
3.density-based clustering		- one student (Elena) presented the previous homework for
other students. It was useful for whom have not done the assigment.
-The number of participants was high
-The instructor monitored the progress of the students as much as possible during the course		-Only few students were engaged in the discussion.
-The presented materials was pretty dense and I guess the students require to get familiar with terminologies and the concepts in advance 		-Follow up with the given assigment in the class to practice the clustering approaches provided in the Jupyer notebook
16
Wed, Apr 17📍On-site10Revised a number of topics that was named by the attendees- good interaction with the studens, lots of questions and interactions
17
Mon, Apr 22💻 Online11
18
Wed, Apr 24📍On-site12Mid Semester CelebrationRecap of concepts previously covered.good interaction
19
Mon, Apr 29Mid Semester PlanningFRANCISCO:
Class Assignments: We've successfully assigned owners to those classes that were lacking them and identified who'll be handling the feedback.

Curriculum Strategy Discussion:
Student Progress: There's a bit of uncertainty on whether all students are fully up to date.
Depth vs. Breadth: We talked about digging deeper into fewer topics rather than skimming through many.
Math vs. Practical Skills: We had a discussion with the following questions: Should we start with the hardcore math or get straight into practical applications? What’s our main goal here? Do we want to ground them in the solid fundamentals of Machine and Deep Learning, or are we aiming for more hands-on skills?
Curriculum Changes: Decided to drop the "DL Timeseries" and some miscellaneous deep learning content in favor of adding two more practical lessons.
Next Steps:
Project Development: We need to figure out how to better develop and integrate projects into the curriculum. Also, how can we better support our students during this crucial phase? This is what we'll focus on next.

ANDREI (Final Project)
1. We will kick-off the Final Project phase on June 5th! During this session we will explain the students what is the Final Project, what is expected from them, how we collaborate, etc. Also we will give them a choice of 3 projects to work on, based on the students preferences for the project topic later we will form groups of 2-3 people (better 3) each who will work on the chosen topic in this group till the very end of semester.
2. Next 4 sessions we will be in Project Preparation mode: here we don't have any strict agenda, groups come, teachers also come. We track attendance and create break-out rooms (virtual or physical) for each group present, teachers will either take couple of groups with the same project topic at a time or just walk in and out of those break out rooms to give all groups fair amount of attention and support. Last project prep session will be also a Demo Day dry-run.
3. Cherry on the top: on June 26th we gonna have a Demo Day :man_dancing: Students will present their work and later the chosen group will present in front of the whole semester cohort. (@Carolina (she/her)
gave way more details in her pitch yesterday - check it out)
CALL TO ACTION
As said, we need 3 project topics: this includes dataset (physical or how to get it: csv. parquet, Kaggle link, etc.), some validated idea (e.g. we are sure that we can predict survivors from Titanic, given Titanic dataset).
We thought that it would be cool to give 3 options:
- Tabular data (regression or classification task)
- NLP case
- Computer Vision case
Boundaries:
- Cases should be not super compute heavy, so students can solve them using quite commodity hardware - we focus on the process.
- Case should be solvable to certain extent using classical approaches (say Sklearn linear regression, HOGs, BoW, etc.)
- Case should be solvable with good quality using DL techniques (say replace above with FCNN regressor, some CNN and RNN)
Minimal success criteria:
- Dataset loaded and preprocessed
- EDA done with some results
- Classical solution present and quality measured
- DL solution present and quality measured
- Presentation with results prepared and pitched
Extra could be (totally up to students to do and decide):
- Deployment as API
- Dockerizing the model
- Building some fancy visualisations with Analytical tools
etc.

Some inspiration for ideas can be taken from our friends in Data Circle
20
Wed, May 1Public Holiday
21
Mon, May 6💻 Online13
22
Wed, May 8📍On-site14AzadLinear Regression using Pytorch
Basic of matrix manipulation
The number of participants was low and we had many absents without a proper excuse
23
Mon, May 13💻 Online15
24
Wed, May 15📍On-site16Olena- coding tasks; working with images - dataset CIFAR10
- low attendance		Same as May 8
25
Mon, May 20Public Holiday
26
Wed, May 22📍On-site17Overall NLP and problems around it. Tokenization, Segmentation, Removing Stop Words, Stemming, Lemmaatization, BOW, Ngrams, TF-IDFAzad explained things vevry clearly and as basic aas possible. Gave very understandable examples and covered all basics of NLP.Attendance was low yet people came to the class put their attentionThere is a homework about applying TF-IDF
27
Mon, May 27💻 Online18
28
Wed, May 29📍On-site19
29
Mon, Jun 3💻 Online20Summary of basic NLP, after that details of embeddings, transformers, laanguage modelsMost of the parts explained very clearly. Live demo sessions was quite intuiitve and interesting.Attendance was fine. Half of the people attended to the class by keeping their cams on and they paid attention. The other half was silent and dark. They showed no participation, therefore we cannot know if they could follow or not.
30
Wed, Jun 5📍On-site21
31
Mon, Jun 10💻 Online22Summary: Project work and doubts from students. Each teacher was in an individual group supporting groupsSome groups split up, let's make sure, everyone can contribute
32
Wed, Jun 12📍On-site23
33
Mon, Jun 17
34
Wed, Jun 19📍On-site24
35
Mon, Jun 24💻 Online25Presentation pitch of various groups, Q & A
36
Wed, Jun 26📍On-site26Demo Day
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100