ECOLE POLYTECHNIQUE

A computer-implemented method of machine-learning. The method includes obtaining a training dataset of 3D models of real-world objects. The method further includes learning, based on the training dataset and on a patch-decomposition of the 3D models of the training dataset, a finite codebook of quantized vectors and a neural network. The neural network comprises a rotation-invariant encoder. The rotation-invariant encoder is configured for rotation-invariant encoding of a patch of a 3D model into a quantized latent vector of the codebook. The neural network further includes a decoder. The decoder is configured for decoding a sequence of quantized latent vectors of the codebook into a 3D model. The sequence corresponds to a patch-decomposition. This constitutes an improved solution for 3D model generation.

A computer-implemented method for determining a quantum phase parameter characterizing a property of a quantum phase of a solid-state material sample, which comprises determining, for a bath site of an impurity-bath model of the solid-state material sample, respective parameters V p and ε p of the bath site, where V p represents a coupling strength of the bath site, and ε p represents an energy of the bath site, configuring a quantum circuit for computing a Green's function of the solid-state material sample corresponding to the impurity-bath model, wherein one or more quantum gates of the quantum circuit are configured based on the determined respective parameters V p and ε p , wherein the quantum circuit is configured to use one or more noisy qubits, wherein respective noise patterns of the noisy qubits are used to represent dissipation of the impurity-bath model; and executing, on a quantum computer, the quantum circuit to determine an estimate of the quantum phase parameter.

A computer-implemented method of applying a machine-learning function preconfigured for taking an input 3D layout and a given noise level, and for predicting an output 3D layout. The function is preconfigured with a conditioning drop-out with respect to at least one layout parameter. The method further comprises obtaining a set of conditioning inputs and, for each conditioning input, determining one or more conditioning candidate 3D layouts and determining a plurality of perturbed conditioning candidate 3D layouts. The method further includes, applying the preconfigured function to each perturbed conditioning candidate, in which the one layout parameter is dropped out, computing reconstruction errors, and averaging the reconstruction errors, thereby obtaining a score. This forms an improved solution for predicting 3D layouts.

The invention relates to a method for characterising the deformability of cells or a portion of cells in a sell sample, the method comprising: culturing the cells on a microstructured plate having microgrooves with a predetermined width and depth for at least partially engaging the nuclei of the cells in one or more microgrooves, at least one portion of the surface of the microgrooves being a cell adhesion surface; measuring a fluorescence signal of the nuclei, which nuclei are pre-treated so as to emit fluorescence radiation; on the basis of the fluorescence signal measured for each nucleus, determining a fluorescence intensity profile and at least one morphological parameter of the nucleus; on the basis of the fluorescence intensity profile and the at least one morphological parameter, determining a deformation class of the nucleus in the depth of the microgrooves.

A plenoptic imaging device for forming an image of the light field of an object, includes a sample holder suitable for holding the object; a main optical assembly, arranged to focus rays of the light field of the object, at a distance z 0 , referred to as the object distance, from the sample holder, which distance is smaller than an object focal distance f 1 from the main optical assembly so as to form an intermediate optical image (I v ) of the object in an intermediate image plane at a distance z 1 from the optical assembly; an assembly for sampling the light field, which assembly is suitable for acquiring spatiodirectional information on the rays forming the virtual image and for forming the image of the light field.

A process for producing nanoclusters of silicon and/or germanium exhibiting a permanent magnetic and/or electric dipole moment for adjusting the work function of materials, for micro- and nano-electronics, for telecommunications, for “nano-ovens”, for organic electronics, for photoelectric devices, for catalytic reactions and for fractionation of water.

A method of quantifying progression of pregnancy, includes a) illuminating a uterine cervix with polarized light having a wavelength comprised between 600 and 1100 nm; b) acquiring a plurality of images of the illuminated uterine cervix through a polarization state analyser in respective analysing states; steps a) and b) being performed either once or a plurality of time using polarized light having different polarization states; c) from the acquired images, computing a depolarization parameter over a region of interest of the uterine cervix; and d) quantifying the progression of the pregnancy from said depolarization parameter. An apparatus, based on a colposcope fitted with a controllable polarization state generator, a controllable polarization state analyser and a suitably-programmed computer, for carrying out such a method. A computer program for carrying out such a method.

The invention relates to a sol formulation which can be used to form a solar-control coating, in particular a coating that blocks UV and NIR radiation, comprising at least: one or more silica-based sol-gel matrix precursors, and doped tungsten bronze nanocrystals which are dispersed uniformly and individually in a protic solvent medium. The invention also relates to a method for forming a solar-control coating at the surface of a support using such a sol formulation, and also to a structure comprising at least one support, preferably a transparent support, having such a solar-control coating on at least one face thereof.

A method implemented by a computer for characterizing at least one observation y of a subject, including the steps of determining a template y characterizing a population of subjects without anomaly, by diffeomorphic deformation, minimizing the cost function J: J(F,A)=K∥y−F( y )−A∥ 2 +λ∥A∥ 1 +Reg(F) where F is a deformation function, A is an anomaly matrix, ∥A∥ 1 is the 1-norm of A, K and λ are predefined constants, Reg is a regularization function, where during minimization, F and A are determined by learning, F providing information on the morphological variability of the subject and A providing information on anomalies in the observation y.

The present invention relates to new specific 6-6 or 5-6 fused bicyclic compounds comprising pyrimidine or pyridine useful in the prevention and/or treatment of infectious diseases. In particular, the present invention relates to a compound of formula (I) wherein: A r1 is a (C 5 -C 11 )arylene or (C 5 -C 11 )heteroarylene group, X is —CH—, —S—, —NR 5 or —N—, Y is —CH—, —NR 5 —S— or —NR 6 —CH 2 —, T is —CH— and Q is —CH— or T is —N— and Q is —CR 10 — or —N—, provided that one or two of X, Q and Y comprise a heteroatom, and with the proviso that, at least one of R 1 , R 2 , and, if present, R 5 or R 6 , contains a group —NH-Alk-NR 3 R 4 . The inventors showed that compounds of formula (I) present an activity against both W2 and 3D7 Plasmodium falciparum strains, an activity against T. brucei brucei but also an activity against SARS-CoV-2 virus, and that they are positive for G4 recognition. The invention also relates to the preparation process and to the therapeutic uses of the compounds of formula (I).