No abstract

No abstract

Human action recognition from skeletal data is an important and active area of research in which the state of the art has not yet achieved near-perfect accuracy on many wellknown datasets. In this paper, we introduce the Distribution of Action Movements Descriptor, a novel action descriptor based on the distribution of the directions of the motions of the joints between frames, over the set of all possible motions in the dataset. The descriptor is computed as a normalized histogram over a set of representative directions of the joints, which are in turn obtained via clustering. While the descriptor is global in the sense that it represents the overall distribution of movement directions of an action, it is able to partially retain its temporal structure by applying a windowing scheme.The descriptor, together with a standard classifier, outperforms several state-of-the-art techniques on many wellknown datasets.

Convolutional neural networks (CNN) offer state-of-the-art performance in various computer vision tasks such as activity recogni tion, face detection, medical image analysis, among others. Many of those tasks need invariance to image transformations (i.e.. rotations, transla tions or scaling). This work proposes a versatile, straightforward and interpretable mea sure to quantify the (in)variance of CNN activations with respect to transformations of the input. Intermediate output values of feature maps and fully connected layers are also analyzed with respect to different input transformations. The technique is applicable to any type of neu ral network and/or transformation. Our technique is validated on ro tation transformations and compared with the relative (in)variance of several networks. More specifically, ResNet, AllConvolutional and VGG architectures were trained on CIFAR10 and MNIST databases with and without rotational data augmentation. Experiments reveal that rotation (in)variance of CNN outputs is class conditional. A distribution analysis also shows that lower layers are the most invariant, which seems to go against previous guidelines that recommend placing invariances near the network output and equivariances near the input.

Feature selection (FS) techniques generally require repeatedly training and evaluating models to assess theimportance of each feature for a particular task. However, due to the increasing size of currently availabledatabases, distributed processing has become a necessity for many tasks. In this context, the Apache SparkML library is one of the most widely used libraries for performing classification and other tasks with largedatasets. Therefore, knowing both the predictive performance and efficiency of its main algorithms beforeapplying a FS technique is crucial to planning computations and saving time. In this work, a comparativestudy of four Spark ML classification algorithms is carried out, statistically measuring execution times andpredictive power based on the number of attributes from a colon cancer database. Results were statistically analyzed, showing that, although Random Forest and Na¨ıve Bayes are the algorithms with the shortest execution times, Support Vector Machine obtains models with the best predictive power. The study of the performance of these algorithms is interesting as they are applied in many different problems, such as classification of pathologies from epigenomic data, image classification, prediction of computer attacks in network security problems, among others.