AI & QC weekly meeting

Europe/Zurich
513/1-024 (CERN)

513/1-024

CERN

1
Show room on map
Videoconference
AI & QC weekly meeting
Zoom Meeting ID
62263502293
Host
Anna Ferrari
Useful links
Join via phone
Zoom URL
    • 2:30 PM 3:00 PM
      Updates 30m
      • Anna Ferrari 1m
        Speaker: Anna Ferrari (CERN)

        updates

        Type*

        3 layer maps D
        platform organization D/T
        video CS4OD D
           
           

        * Type: T = to do / D = done/ O = others

      • José Cabrero Hogueras 1m
        Speaker: Jose Cabrero Holgueras (University Carlos III (ES))

        updates

        Type*

        Matrix homomorphic operations abstraction library T/D
           
           
           
           

        * Type: T = to do / D = done/ O = others

      • Vasileios Belis 1m
        Speaker: Vasileios Belis (ETH Zurich (CH))

        updates

        Type*

        Preparation for first teaching session  D
        Talk on QSIT/Quantum center ETH Zurich  T/D
        Higgs classification + AE repo finalization  T/D
        1-class SVM class inheritance from QuantumKernel  T/D
           

        * Type: T = to do / D = done/ O = others

      • Marco Rossi 1m
        Speaker: Marco Rossi (CERN)

        updates

        Type*

        Benchmark slicing on enlarged dataset D
        Giving lectures at university

        D

        Pileup subtraction with reinforcement learning T/D

        * Type: T = to do / D = done/ O = others

      • Su Yeon Chang 1m
        Speaker: Su Yeon Chang (EPFL - Ecole Polytechnique Federale Lausanne (CH))

        updates

        Type*

        Implement& Test quantum classifier for multiclass  T/D 
        Test 1d qgan with full noise model  D

        * Type: T = to do / D = done/ O = others

      • Florian Rehm 1m
        Speaker: Florian Rehm (Rheinisch Westfaelische Tech. Hoch. (DE))

        Finishing university lectures.
        Evaluate quantized model.
        Quantum computing parameter optimization.

      • Gabriele Morello 1m
        Speaker: Gabriele Morello (Politecnico di Torino (IT))

        updates

        Type*

        rewriting encryption according to protobuf structure

        T/D

        run openfl on a single machine (useless for my project) D
        run openfl on more than one machine  T
           
           

        * Type: T = to do / D = done/ O = others

      • Renato Cardoso 1m
        Speaker: Renato Paulo Da Costa Cardoso (Universidade de Lisboa (PT))

        updates

        Type*

        Finish intel Progressive Gan run on GPUs D
        Testing different approaches to color consistency problem with AE D/T
           
           
           

        * Type: T = to do / D = done/ O = others

      • Denis-Partick Odagiu 1m
        Speaker: Denis-Patrick Odagiu

        updates

        Type*

        Optimized the classical classifier lr and batch size.
        Could not remove the 'wild oscillations' in the loss of the last training batch,
        but understood why they are happening and they are not unnatural for such a network.
        D/T
        Looked at the raw trees for the data. Found variables jets_MatchFlag_1...7. 
        These might contain information on if the jet is coming from Higgs or not,
        need to confirm with Christina.
        D
        Restructured the ae code to rely more on inheritance from a base vanilla class.
        Makes adding variations on the ae easier.
        D
        Adding an SVM instead of the DNN to perform the classification task, 
        to check if this performs better.
        T/D
        While only the classifier branch was active, I trained the ae + classifier with the optimized
        hyperparameters and calculated the AUC in the 'latent' space and at the end of the network.
        I obtained a better AUC than 0.71... For both.
        D

        * Type: T = to do / D = done/ O = others

      • Diamantis Patsidis 1m
        Speaker: Diamantis Patsidis

        updates

        Type*

        create library for the docker interface   D 

        add support for alpine/fedora operating

        systems in library

         

         D

         

        implemented input security checks

         D

        add functionality for extending pre-built

        docker images

         

         D

         

        run library tests

         T

        * Type: T = to do / D = done/ O = others

         

         

      • Hanyul Ryu 1m
        Speaker: Hanyul Ryu (The Seoul National University Bundang Hospital (KR))

        updates

        Type*

        semi-fin QKD simulator connection to WAS T/D
        HE image processing and QKD simulator connection T
        Processing of visa document O
           
           

        * Type: T = to do / D = done/ O = others

      • Kristina Jaruskova 1m
        Speaker: Kristina Jaruskova (CERN)

        updates

        Type*

        Changes in validation function for Conv2D D
        Meeting with Emma and Adel from Intel (intro, discussing possible cooperation) D
        Prepare materials for Joseph D
        AdaGAN T/D
           

        * Type: T = to do / D = done/ O = others

      • Oriel Kiss 1m
        Speaker: Oriel Orphee Moira Kiss (Universite de Geneve (CH))

        updates

        Type*

        born machine with multiple output (3)  
           
           
           
           

        * Type: T = to do / D = done/ O = others

      • Yayan Tan 1m
        Speaker: Yayan Tan (University of Cambridge)

        updates

        Type*

        Compute the magnitude data and drow boxplot D
        Finding the frequency range for filtering  T/D
        Boxplot and K-Means clustering for data after filtering T
        Compute K-means clustering for raw data T
           

        * Type: T = to do / D = done/ O = others

    • 3:00 PM 3:30 PM
      Technical presentation and references 30m
      Speaker: Hanyul Ryu (The Seoul National University Bundang Hospital (KR))