it misses the V0 content ! you have to insert everything which was in the V0 in the v1
think about the structure of the document :
Introduction
Contexte du projet
Applications : modèle réduit pour des problèmes en grande dimension
Objectives :
un objectif global : à partir de données, trouver une structure algébrique qui explique ces données.
sous objectif 1 : identifier la meilleure facon
sous objectif 2 : implémenter un cas 1D
sous objectif 3 : étendre à des dimensions plus grandes
First cases : «almost affine» manifolds
4.1 Motivation (pourquoi on a choisi ce cas test et on explique )
4.2 Numerical experiments
4.2.1 One dim
4.2.2. 2D+
General manifolds
5.1 Setting a general distance on M
5.2 Numerical experiments
5.2.1 One dim case (§ training et § validation )
5.2.2 2D case (idem)
Conclusions
Reformulate some epxressions : «Developed Code» -> «Our contribution» , «Analysis and interpretation» --> «Numerical experiments» , «Dataset» est-ce pour l'entrainement ? ou le test ? préciser la nature et le nombre de points que vous avez choisi
add some figures to explain every objects which is trained
don't forget the sum $\sum_{i,j}$ !! in the losses
Explain in detail the validation step : what is the test dataset ? how do you validate your results ? etc
Talk about the Unicity of the minimization problem : can we expect one solution ? or an ininifite ? investigate this with random initialization of the minimization...
The report in the V1 release has some problems :
it misses the V0 content ! you have to insert everything which was in the V0 in the v1
think about the structure of the document :
4.2.1 One dim 4.2.2. 2D+
Reformulate some epxressions : «Developed Code» -> «Our contribution» , «Analysis and interpretation» --> «Numerical experiments» , «Dataset» est-ce pour l'entrainement ? ou le test ? préciser la nature et le nombre de points que vous avez choisi
add some figures to explain every objects which is trained
don't forget the sum $\sum_{i,j}$ !! in the losses
Explain in detail the validation step : what is the test dataset ? how do you validate your results ? etc
Talk about the Unicity of the minimization problem : can we expect one solution ? or an ininifite ? investigate this with random initialization of the minimization...
Fig 3 : expliquer ce que l'on voit