Syllabus del corso
Contenuti sintetici
The course introduces general aspects of Machine Learning and Deep Learning and
discusses the most common architectures used, e.g., for Computing Vision and Natural
Language Processing. These lectures are complemented by hands-on sessions, where
different architectures are used to solve a real-life classification problem typical of Physics
data analysis. The hands-on session is based on examples of LHC data analysis coded in
Keras/TensorFlow and Pytorch, running on Jupiter notebooks via Colab.
Programma esteso
DAY 1 [2h]:
Lecture 1: Introduction to Deep Learning
Hands-on tutorial: Fully connected network for jet classification at the LHC
Day 2 [2h]:
Lecture 2: Computing vision with Convolutional Neural Networks
Hands-on tutorial: CNN for jet classification at the LHC
Day 3 [2h]:
Lecture 3: Recurrent Neural Networks
Hands-on tutorial: RNN for jet classification at the LHC
Day 4 [2h]:
Lecture 4: Graph Neural Networks
Hands-on tutorial: Jet classification at the LHC with Graph Networks
Day 5 [2h]:
Lecture 5: Unsupervised Learning and Anomaly Detection
Hands-on tutorial: Anomalous jet detection at the LHC
Modalità didattica
1 CFU, 10 ore.
Periodo di erogazione dell'insegnamento
3, 4, 5, 6, 7 maggio 2021 h. 14:00-16:00.
Contents
The course introduces general aspects of Machine Learning and Deep Learning and
discusses the most common architectures used, e.g., for Computing Vision and Natural
Language Processing. These lectures are complemented by hands-on sessions, where
different architectures are used to solve a real-life classification problem typical of Physics
data analysis. The hands-on session is based on examples of LHC data analysis coded in
Keras/TensorFlow and Pytorch, running on Jupiter notebooks via Colab.
Detailed program
DAY 1 [2h]:
Lecture 1: Introduction to Deep Learning
Hands-on tutorial: Fully connected network for jet classification at the LHC
Day 2 [2h]:
Lecture 2: Computing vision with Convolutional Neural Networks
Hands-on tutorial: CNN for jet classification at the LHC
Day 3 [2h]:
Lecture 3: Recurrent Neural Networks
Hands-on tutorial: RNN for jet classification at the LHC
Day 4 [2h]:
Lecture 4: Graph Neural Networks
Hands-on tutorial: Jet classification at the LHC with Graph Networks
Day 5 [2h]:
Lecture 5: Unsupervised Learning and Anomaly Detection
Hands-on tutorial: Anomalous jet detection at the LHC
Teaching form
1 CFU, 10 hours.
Semester
May 3, 4, 5, 6, 7 2021 h. 14:00-16:00.