MLHEP 2015 summer school
27 Aug 2015, 08:00
→
30 Aug 2015, 23:00
Europe/Moscow
Academic University
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Description
Summer school on Machine Learning in High Energy Physics
http://www.hse.ru/mlhep2015/
Kaggle COMET competition
MLHEP github repo
Thursday 27 August
Thu 27 Aug
Fri 28 Aug
Sat 29 Aug
Sun 30 Aug
08:00
→
08:15
Registration
08:15
→
09:00
breakfast
45m
09:00
→
09:35
Welcome
204
204
Academic University
Convener
:
Andrey Ustyuzhanin
(
Yandex School of Data Analysis (RU)
)
09:00
Welcome to MLHEP
20m
2015-08_MLHEP_Welcome.pdf
09:20
MLHEP Kaggle competition
15m
Speaker
:
Nikita Kazeev
(
Yandex School of Data Analysis (RU)
)
mlhep2015_kaggle_slides.pdf
09:35
→
09:40
break
5m
09:40
→
12:40
Advanced track: lectures
206
206
Academic University
Convener
:
Dr
Victor Kitov
(
MSU
)
09:40
Reminder about major algorithms. Advanced aspects of their use. Model evaluation.
1h 20m
Introduction. Model evaluation.pdf
11:00
coffee break
20m
11:20
Feature selection. Regularization.
1h 20m
Feature selection.pdf
09:40
→
12:40
Introductory track: lectures
204
204
Academic University
Convener
:
Alexey Rogozhnikov
09:40
Intro: General pipeline, ML at a glance
1h 20m
lecture1.pdf
11:00
coffee break
20m
11:20
Intro: kNN, basic overfitting, roc curve, logistic regression
1h 20m
12:40
→
13:45
lunch
1h 5m
13:45
→
17:10
Advanced track: seminars
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Tatiana Likhomanenko
(
National Research Centre Kurchatov Institute (RU)
)
13:45
MLHEP introduction. ML vs MLHEP: correlation and agreement restrictions
1h 30m
Speaker
:
Tatiana Likhomanenko
(
National Research Centre Kurchatov Institute (RU)
)
seminar1_part1.pdf
15:15
coffee break
25m
15:40
sPlot technique. ML on sPlot data.
1h 30m
seminar1_part2.pdf
13:45
→
17:10
Introductory track: seminars
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Nikita Kazeev
(
Yandex School of Data Analysis (RU)
)
13:45
Introduction, course technicalities
1h 30m
1.1 Technicalities and principles.ipynb
1.2 Tools numpy pandas matplotlib - seminar.ipynb
15:15
coffee break
25m
15:40
NumPy, Pandas, Matplotlib
1h 30m
17:10
→
17:20
break
10m
17:20
→
19:00
Applying Machine Learning to LHC triggers optimization
1h 40m
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Speaker
:
Mika Anton Vesterinen
(
Ruprecht-Karls-Universitaet Heidelberg (DE)
)
Mika_MLHEP.pdf
19:00
→
20:30
Welcome dinner
1h 30m
Friday 28 August
Thu 27 Aug
Fri 28 Aug
Sat 29 Aug
Sun 30 Aug
08:15
→
09:30
breakfast
1h 15m
09:30
→
12:30
Advanced track: lectures #2
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Dr
Victor Kitov
(
MSU
)
09:30
model ensembling #1
1h 20m
10:50
coffee break
20m
11:10
model ensembling #2
1h 20m
09:30
→
12:30
Introductory track: lectures #2
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Alexey Rogozhnikov
09:30
neural networks
1h 20m
lecture2.pdf
10:50
coffee break
20m
11:10
decision tree, regression tree
1h 20m
12:30
→
13:45
lunch
1h 15m
13:45
→
17:10
Advanced track: seminars #2
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Tatiana Likhomanenko
(
National Research Centre Kurchatov Institute (RU)
)
13:45
Brave new boosting world: flatness boosting
1h 30m
seminar2.pdf
15:15
coffee break
25m
15:40
Brave new boosting world: reweighing boosting. Iterative learning
1h 30m
13:45
→
17:10
Introductory track: seminars #2
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Nikita Kazeev
(
Yandex School of Data Analysis (RU)
)
13:45
Overfitting, Cross-validation
1h 30m
2.1. Basic ML, validation, feature engineering.ipynb
15:15
coffee break
25m
15:40
Sklearn: KNN, decision tree, logistic regression
1h 30m
17:10
→
17:30
break
20m
17:30
→
19:10
Challenges of searching for physics signatures with a large amount of background
1h 40m
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Speaker
:
Patrick Haworth Owen
(
Imperial College Sci., Tech. & Med. (GB)
)
Patrick_Machine_Learning_in_LHCb.pdf
Saturday 29 August
Thu 27 Aug
Fri 28 Aug
Sat 29 Aug
Sun 30 Aug
08:15
→
09:30
breakfast
1h 15m
09:30
→
12:30
Advanced track: lectures #3
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Dr
Victor Kitov
(
MSU
)
09:30
Linear dimensionality reduction.
1h 20m
Dimensionality reduction.pdf
10:50
coffee break
20m
11:10
Non-linear dimensionality reduction. Kernel trick. Common kernels.
1h 20m
09:30
→
12:30
Introductory track: lectures #3
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Alexey Rogozhnikov
09:30
Ensembles
1h 20m
lecture3.pdf
10:50
coffee break
20m
11:10
Gradient boosting; boosting to uniformity.
1h 20m
12:30
→
13:45
lunch
1h 15m
13:45
→
17:10
Advanced track: seminars #3
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Tatiana Likhomanenko
(
National Research Centre Kurchatov Institute (RU)
)
13:45
Meta Algorithms Applications in HEP.
1h 30m
seminar3_part1.pdf
15:15
coffee break
25m
15:40
Hypotheses testing.
1h 30m
seminar3_part2.pdf
13:45
→
17:10
Introductory track: seminars #3
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Nikita Kazeev
(
Yandex School of Data Analysis (RU)
)
13:45
Ensemble algorithms: random forest, gradient boosting
1h 30m
3.1 Ensemble methods - seminar.ipynb
COMET_gradient_boosting_gauss_opt.csv
15:15
coffee break
25m
15:40
Hyperparam Optimization
1h 30m
3.2 Hyperparameters optimzation - seminar.ipynb
17:10
→
17:30
break
20m
17:30
→
19:10
MVA vs cut-and-count techniques in the SM model processes searches and measurements
1h 40m
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Speaker
:
Lesya Shchutska
(
University of Florida (US)
)
SUSY_mva.pdf
Sunday 30 August
Thu 27 Aug
Fri 28 Aug
Sat 29 Aug
Sun 30 Aug
08:15
→
09:00
breakfast
45m
09:00
→
12:00
Advanced track: lectures #4
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Dr
Victor Kitov
(
MSU
)
09:00
Kernelized algorithms: SVM, regression, K-NN, PCA and their properties.
1h 20m
Kernel trick. Deep learning.pdf
10:20
coffee break
20m
10:40
Deep learning. Different kinds of learning.
1h 20m
Kernel trick. Deep learning.pdf
Learning Deep Architectures for AI - Bengio.pdf
09:00
→
12:00
Introductory track: lectures #4
204
204
Academic University
Convener
:
Alexey Rogozhnikov
09:00
GBDT tuning, reweighting, testing hypotheses, gaussian processes for optimization
1h 20m
lecture4.pdf
10:20
coffee break
20m
10:40
Unsupervised learning, Latent Dirichlet Allocation, PCA. Fast predictions and pruning
1h 20m
12:00
→
13:00
lunch
1h
13:00
→
16:15
Advanced track: seminars #4
206
206
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Tatiana Likhomanenko
(
National Research Centre Kurchatov Institute (RU)
)
13:00
Summary: HEP ML analysis sketch, from data to discovery.
1h 30m
seminar4_summary.pdf
14:30
coffee break
15m
14:45
My own first NN using theano.
1h 30m
13:00
→
16:15
Introductory track: seminars #4
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Convener
:
Nikita Kazeev
(
Yandex School of Data Analysis (RU)
)
13:00
PCA, RBM and alike on MNIST
1h 30m
4.1 Generative models - seminar.ipynb
14:30
coffee break
15m
14:45
Clustering
1h 30m
4.2. Clustering - seminar.ipynb
16:15
→
16:20
break
5m
16:20
→
18:00
Multivariate techniques in the Higgs search
1h 40m
204
204
Academic University
Khlopin st, 8, Saint Petersburg, Russia
Speaker
:
Josh Bendavid
(
CERN
)
mlhep-Aug30-2015.pdf
18:00
→
19:00
transfer to the boat
1h
19:00
→
22:00
channel cruise + closing dinner + awards
3h
22:00
→
23:00
transfer to the hotel
1h