Skip to content

Homeworks and projects done during the Master Mathematics, Learning, Vision (MVA) at ENS Paris-Saclay.

Notifications You must be signed in to change notification settings

EGhamgui/MVA-Courses

Repository files navigation

Assignments and Projects Realized During the Master Mathématiques, Vision, Apprentissage (MVA) at ENS Paris-Saclay

First Semester

Convex Optimization (ALEXANDRE D'ASPREMONT)

  • HM1: Convexity, Conjugate Function
  • HM2: Duality
  • HM3: LASSO

Computational Statistics (STÉPHANIE ALLASSONIÈRE)

  • HM1: Markov Chains - Stochastic Gradient Descent
  • HM2: Expectation-Maximisation Algorithm – Importance Sampling
  • HM3: Hasting-Metropolis (and Gibbs) Samplers
  • HM4: Metropolis-Hastings Algorithm

Image Denoising (JEAN-MICHEL MOREL, GABRIELE FACCIOLO, PABLO ARIAS)

  • HM1: Ponomarenko Algorithm, Multi-Scale DCT
  • HM2: BM3D, Non-Local Means
  • HM3: Patch Similarity, Non-Local Bayes
  • HM4: EPLL, Zoran-Weiss Gaussians Mixture Model
  • HM5: Introduction to CNN Denoising
  • HM6: A Deeper Understanding of CNN Denoising
  • HM7: Noise to Noise
  • HM8: Different Denoising Architectures

Medical Image Analysis (HERVÉ DELINGETTE, XAVIER PENNEC)

Paper Summary: Boundary Loss for Highly Unbalanced Segmentation
Reference: https://proceedings.mlr.press/v102/kervadec19a.html

Object Recognition and Computer Vision (IVAN LPATEV, JEAN PONCE, CORDELIA SCHMID, JOSEF SIVIC)

Topological Data Analysis for Imaging and Machine Learning (FRÉDÉRIC CHAZAL, JULIEN TIERNY)

Second Semester

Deep Learning in Medical Imaging (OLIVIER COLLIOT, MARIA VAKALOPOULOU)

Deep Learning in Practice (GUILLAUME CHARPIAT)

  • HM1: Hyperparameters Tuning
  • HM2: Grad-CAM
  • HM3: Graph Neural Networks
  • HM4: Small Data: Weak Supervision, Transfer, and Incorporation of Priors
  • HM5: Koopman Decomposition, Duffing Oscillator
  • HM6: Generative Models

Graphs in Machine Learning (DANIELE CALANDRIELLO)

  • HM1: Spectral Clustering
  • HM2: Semi-Supervised Learning (SSL)
  • HM3: Introduction to Graph Neural Nets with JAX/Jraph

Time Series Analysis (LAURENT OUDRE)

  • HM1: Convolutional Dictionary Learning (CDL), Dynamic Time Warping (DTW)
  • HM2: ARIMA Process, Sparse Coding
  • HM3: Change-Point Detection, Wavelet Transform for Graph Signals
  • Project: Feature Selection Methods
    Reference: https://dl.acm.org/doi/pdf/10.1145/3136625

Releases

No releases published

Packages

No packages published