Quantum Dots are semiconductor nanocrystals whose diameter is in the range of 2-10 nm, corre- sponding to 10 to 50 atoms in diameter and a total of 100 to 100,000 atoms within the quantum dot volume. Many types of quantum dot emit light of specific frequencies if electricity or light is applied to them, and these frequencies can be precisely tuned by changing the dots’ size, shape and material, giving rise to many applica- tions. Because of their high tunable properties, quantum dots are of wide interest. It finds its applications in nanotechnology, medical imaging, transistors, solar cells, LED’s, diode lasers, quantum computing, etc. With this project, we intend to further understand and study the properties of quantum dots by using atomic force microscopy.
Cet article étudie deux méthodes utilisées dans le cadre du transport humanitaire en cas de crise (désastre, épidémie...). Le Covering Tour Problem se focalise sur l'équité de distribution des vivres, alors que le Capacitated Vehicle Routing Problem se concentre sur l'urgence de la distribution. Nous proposons une nouvelle approche mélangeant ces deux approches pour former une solution à la fois équitable et rapide. Ce article a été rédigé dans le cadre du TER 2014-2015.
Dimitry Berardi, Abdelwahab Heba, Boris Terooatea, Maël Valais
Although the analysis of data is a task that has gained the interest of the statistical community in recent years and whose familiarity with the statistical computing environment, they encourage the current statistical community (to students and teachers of the area) to complete statistical analysis reproducible by means of the tool R. However for years there has been a gap between the calculation of matrices on a large scale and the term "big data", in this work the Normalized Cut algorithm for images is applied. Despite the expected, the R environment to do image analysis is poorly, in comparison with other computing platforms such as the Python language or with specialized software such as OpenCV.
Being well known the absence of such function, in this work we share an implementation of the Normalized Cut algorithm in the R environment with extensions to programs and processes performed in C ++, to provide the user with a friendly interface in R to segment images. The article concludes by evaluating the current implementation and looking for ways to generalize the implementation for a large scale context and reuse the developed code.
Key words: Normaliced Cut, image segmentation, Lanczos algorithm, eigenvalues and eigenvectors, graphs, similarity matrix, R (the statistical computing environment), open source, large scale and big data.
The Observatory of Public Spending (or ODP, in Portuguese) is a special unit of Brazil's Ministry of Transparency, Monitoring and Office of the Comptroller-General (or CGU, in Portuguese) responsible for monitoring public spending and gathering managerial and audit information to support the work of CGU internal auditors. One of the most important themes monitored by this unit is Public Procurements and Government Suppliers which have won these procurement processes. Image analysis of many of these suppliers headquarters revealed suspicious landscapes, such as rural areas, isolated places or slums. These landscapes could be an indication of fake suppliers with poor capacity of delivering public goods and services. However, checking thousands of landscapes in order to find these fake suppliers would be a very expensive task. Our objective then is to discover what are the possible groups of scenes involving government suppliers, given that these images were not previously labeled, as automatically as possible. For that reason, we used Places CNN, a pretrained convolutional neural network for scene recognition presented by Zhou et al., which was trained on 205 scene categories with 2.5 million images, for scene recognition on Brazilian Government Suppliers.
Comprensión de un estudio realizado en la mica-epoxi para placas de circuitos. El estudio consistió en pruebas de resistencia para medir el desgaste en el tiempo del material y así determinar su tiempo de vida aproximado.
In this paper, we evaluate a baseline word embedding model for a set of clinical notes derived from patient records. For our baseline, we extract features for this embedding using the Word2Vec module from the gensim package. We also build two models, a word2vec skipgram model with negative sampling and a positive point-wise mutual information (PPMI) model by training on the processed clinical notes. Our evaluation shows that both the PPMI and the skipgram models show improved results for medically-related terms when compared with the baseline model. PPMI shows the best result out of all three models.
Paper presented at ICCV 2019.
This paper targets the task with discrete and periodic
class labels (e.g., pose/orientation estimation) in the context of deep learning. The commonly used cross-entropy or
regression loss is not well matched to this problem as they
ignore the periodic nature of the labels and the class similarity, or assume labels are continuous value. We propose to
incorporate inter-class correlations in a Wasserstein training framework by pre-defining (i.e., using arc length of a
circle) or adaptively learning the ground metric. We extend
the ground metric as a linear, convex or concave increasing
function w.r.t. arc length from an optimization perspective.
We also propose to construct the conservative target labels
which model the inlier and outlier noises using a wrapped
unimodal-uniform mixture distribution. Unlike the one-hot
setting, the conservative label makes the computation of
Wasserstein distance more challenging. We systematically
conclude the practical closed-form solution of Wasserstein
distance for pose data with either one-hot or conservative
target label. We evaluate our method on head, body, vehicle and 3D object pose benchmarks with exhaustive ablation studies. The Wasserstein loss obtaining superior performance over the current methods, especially using convex mapping function for ground metric, conservative label,
and closed-form solution.
Xiaofeng Liu, Yang Zou, Tong Che, Peng Ding, Ping Jia, Jane You, B.V.K. Vijaya Kumar