This work was undertaken to analyze the ripening process of avocados variety Hass (Persea americana Mill.) by image processing (IP) methodology. A set of avocados (10 samples) was used to follow the changes in image features during ripening by applying a computer vision system, extracting color and textural parameters. Other 16 avocados were used to evaluate the firmness and mass loss. Three maturity stages of avocados were established, and a classification was obtained by applying principal component analysis and k-nearest neighbor algorithm. During the ripening process (12 days), avocado firmness decreased from 75.43 to 2.63 N, while skin color values kept invariable during 6 days; after that, a decrement in the peel green color (a*) was observed (−9.68 to 2.32). Image features showed that during ripening the color parameters (L*, a*, and b*), entropy (4.29 to 4.00), angular second moment (0.287 to 0.360), and fractal dimension (2.58 to 2.44) had a similar path as compared to mass loss, a*, and firmness ripening parameters, respectively. Relationships between image features and ripening parameters were obtained. The parameter a* was the most useful digital feature to establish an acceptable percentage of avocado classification (>80%) in three different maturity stages found. Results obtained by means of IP could be useful to evaluate, at laboratory level, the ripening process of the avocados.

No abstract

In this work we propose to solve the author verification problem using a semantic space model through Latent Dirichlet Allocation (LDA). We experiment with the corpus used in the author identification tasks at PAN 2014 and PAN 2015. These datasets consist of subsets in the following languages: English, Spanish, Dutch and Greek. Each problem contained in these corpora is formed by one to five known documents which were written by one author and one unknown document. The task is to predict whether the unknown document was written by the author who wrote the known documents. We processed the documents in the dataset and captured the fingerprint of authors by generating a probabilistic distribution of words in the documents. In PAN 2015 classification, we achieved 81.6%, 75.4%, 74.1%, 67.1% accuracy for each English, Spanish, Dutch and Greek subset respectively. In particular for the English subset, we outreached the best result reported in both competitions.

No abstract

Abstract-In this paper, we present a new algorithm, namely, a micro artificial immune system (Micro-AIS) based on the Clonal Selection Theory for solving numerical optimization problems. For our study, we consider the algorithm CLONALG, a widely used artificial immune system. During the process of cloning, CLONALG greatly increases the size of its population. We propose a version with reduced population. Our hypothesis is that reducing the number of individuals in a population will decrease the number of evaluations of the objective function, increasing the speed of convergence and reducing the use of data memory. Our proposal uses a population of 5 individuals (antibodies), from which only 15 clones are obtained. In the maturation stage of the clones, two simple and fast mutation operators are used in a nominal convergence that works together with a reinitialization process to preserve the diversity. To validate our algorithm, we use a set of test functions taken from the specialized literature to compare our approach with the standard version of CLONALG. The same method can be applied in many other problems, for example, in text processing.

No abstract

No abstract

This work presents a model based on Deep Neural Networks for the prediction of apparent personality. It can quantify personality traits with the Five-Factor model (Big Five) from a Portrait image. In order to evaluate the effectiveness of this approach, a new corpus of 30,935 portraits with their associated personality trait was extracted from an existing resource of videos (First Impressions, ChaLearn) tagged with redundant pairwise comparisons to ensure consistency. We propose several models using Convolutional Neural Networks to automatically extract features from a portrait that are indicators of personality traits; then the models classify these characteristics into a binary class for each Big Five factor: openness to experience (O), conscientiousness (C), extraversion (E), agreeableness (A), and neuroticism (N). In addition, we experiment with feature encoding and transfer learning to enrich the representation of images with additional untagged portraits (∼45,000 and ∼200M), reaching a percentage of accuracy within the state of the art (albeit not directly comparable), obtaining 65.86% as a classifier averaging the 5 factors (O=61.48%, C=69.56%, E=73.23%, A=60.68%, N=64.35%). Compared to human judgment (mean accuracy of 56.66%), the model obtained higher average performance and higher accuracy in 4 of the 5 factors of the Big Five model. In addition, in comparison with the state of the art this model shows several advantages: (1) it requires only a single portrait to make the prediction, being this a non-invasive and easily accessible resource (e.g. selfies) (2) the extraction of features from the portrait is done automatically, (3) a single model performs the extraction of characteristics and classification.