Browsing by Autor "Cristian Lopez"

Now showing 1 - 3 of 3

Approximate nearest neighbors by deep hashing on large-scale search: Comparison of representations and retrieval performance
(2017) Alexander Ocsa; Jose Luis Huillca; Ricardo Coronado; Oscar Quispe; Carlos Arbieto; Cristian Lopez
The growing volume of data and its increasing complexity require even more efficient and faster information retrieval techniques. Approximate nearest neighbor search algorithms based on hashing were proposed to query high-dimensional datasets due to its high retrieval speed and low storage cost. Recent studies promote the use of Convolutional Neural Network (CNN) with hashing techniques to improve the search accuracy. However, there are challenges to solve in order to find a practical and efficient solution to index CNN features, such as the need for a heavy training process to achieve accurate query results and the critical dependency on data-parameters. In this work we execute exhaustive experiments in order to compare recent methods that are able to produces a better representation of the data space with a less computational cost for a better accuracy by computing the best data-parameter values for optimal sub-space projection exploring the correlations among CNN feature attributes using fractal theory. We give an overview of these different techniques and present our comparative experiments for data representation and retrieval performance.
Efficient approach for interest points detection in non-rigid shapes
(2015) Cristian Lopez; Luciano Arnaldo Romero Calla; Lizeth Joseline Fuentes Perez
Due to the increasing amount of data and the reduction of costs in 3D data acquisition devices, there has been a growing interest, in developing efficient and robust feature extraction algorithms for 3D shapes, invariants to isometric, topological and noise changes, among others. One of the key tasks for feature extraction in 3D shapes is the interest points detection; where interest points are salient structures, which can be used, instead of the whole object. In this research, we present a new approach to detect interest points in 3D shapes by analyzing the triangles that compose the mesh which represent the shape, in different way to other algorithms more complex such as Harris 3D or HKS. Our results and experiments of repeatability, confirm that our algorithm is stable and robust, in addition, the computational complexity is O(n log n), where n represents the number of faces of the mesh.
GAF-CNN-LSTM for Multivariate Time- Series Images Forecasting
(Centre National de la Recherche Scientifique, 2019) Edson F. Luque; Cristian Lopez
Forecasting multivariate time series is challenging for a whole host of reasons not limited to problem features such as having multiple input variables, time series preparation, and the need to perform the same type of prediction for multiple physical sites. Although the literature on time series forecasting is focused on 1D signals. We use the Gramian Angular Fields (GAFs) to encode time series into 2D texture images, later take advantage of the deep CNN-LSTM architecture where LSTM uses a CNN as front end. Thus, we propose a novel unified framework for forecasting multivariate time series using a way to encode time series as images. Preliminary experimental results on the UEA multivariate time series forecasting archive, demonstrate competitive forecast accuracy (RMSE and MAPE) of the proposed approach, compared to the existing deep approaches as LSTM, CRNN, 1D-MTCNN.