Models, code, and papers for "Thiago M":

Image clustering is an important but challenging task in machine learning. As in most image processing areas, the latest improvements came from models based on the deep learning approach. However, classical deep learning methods have problems to deal with spatial image transformations like scale and rotation. In this paper, we propose the use of visual attention techniques to reduce this problem in image clustering methods. We evaluate the combination of a deep image clustering model called Deep Adaptive Clustering (DAC) with the Visual Spatial Transformer Networks (STN). The proposed model is evaluated in the datasets MNIST and FashionMNIST and outperformed the baseline model in experiments.

Non-parametric approaches for analyzing network data based on exchangeable graph models (ExGM) have recently gained interest. The key object that defines an ExGM is often referred to as a graphon. This non-parametric perspective on network modeling poses challenging questions on how to make inference on the graphon underlying observed network data. In this paper, we propose a computationally efficient procedure to estimate a graphon from a set of observed networks generated from it. This procedure is based on a stochastic blockmodel approximation (SBA) of the graphon. We show that, by approximating the graphon with a stochastic block model, the graphon can be consistently estimated, that is, the estimation error vanishes as the size of the graph approaches infinity.

Dynamic regressor selection (DRS) systems work by selecting the most competent regressors from an ensemble to estimate the target value of a given test pattern. This competence is usually quantified using the performance of the regressors in local regions of the feature space around the test pattern. However, choosing the best measure to calculate the level of competence correctly is not straightforward. The literature of dynamic classifier selection presents a wide variety of competence measures, which cannot be used or adapted for DRS. In this paper, we review eight measures used with regression problems, and adapt them to test the performance of the DRS algorithms found in the literature. Such measures are extracted from a local region of the feature space around the test pattern, called region of competence, therefore competence measures.To better compare the competence measures, we perform a set of comprehensive experiments of 15 regression datasets. Three DRS systems were compared against individual regressor and static systems that use the Mean and the Median to combine the outputs of the regressors from the ensemble. The DRS systems were assessed varying the competence measures. Our results show that DRS systems outperform individual regressors and static systems but the choice of the competence measure is problem-dependent.

Many popular Bayesian nonparametric priors can be characterized in terms of exchangeable species sampling sequences. However, in some applications, exchangeability may not be appropriate. We introduce a {novel and probabilistically coherent family of non-exchangeable species sampling sequences characterized by a tractable predictive probability function with weights driven by a sequence of independent Beta random variables. We compare their theoretical clustering properties with those of the Dirichlet Process and the two parameters Poisson-Dirichlet process. The proposed construction provides a complete characterization of the joint process, differently from existing work. We then propose the use of such process as prior distribution in a hierarchical Bayes modeling framework, and we describe a Markov Chain Monte Carlo sampler for posterior inference. We evaluate the performance of the prior and the robustness of the resulting inference in a simulation study, providing a comparison with popular Dirichlet Processes mixtures and Hidden Markov Models. Finally, we develop an application to the detection of chromosomal aberrations in breast cancer by leveraging array CGH data.

Bayesian Optimization (BO) is a framework for black-box optimization that is especially suitable for expensive cost functions. Among the main parts of a BO algorithm, the acquisition function is of fundamental importance, since it guides the optimization algorithm by translating the uncertainty of the regression model in a utility measure for each point to be evaluated. Considering such aspect, selection and design of acquisition functions are one of the most popular research topics in BO. Since no single acquisition function was proved to have better performance in all tasks, a well-established approach consists of selecting different acquisition functions along the iterations of a BO execution. In such an approach, the GP-Hedge algorithm is a widely used option given its simplicity and good performance. Despite its success in various applications, GP-Hedge shows an undesirable characteristic of accounting on all past performance measures of each acquisition function to select the next function to be used. In this case, good or bad values obtained in an initial iteration may impact the choice of the acquisition function for the rest of the algorithm. This fact may induce a dominant behavior of an acquisition function and impact the final performance of the method. Aiming to overcome such limitation, in this work we propose a variant of GP-Hedge, named No-PASt-BO, that reduce the influence of far past evaluations. Moreover, our method presents a built-in normalization that avoids the functions in the portfolio to have similar probabilities, thus improving the exploration. The obtained results on both synthetic and real-world optimization tasks indicate that No-PASt-BO presents competitive performance and always outperforms GP-Hedge.

Deep learning techniques have enabled the emergence of state-of-the-art models to address object detection tasks. However, these techniques are data-driven, delegating the accuracy to the training dataset which must resemble the images in the target task. The acquisition of a dataset involves annotating images, an arduous and expensive process, generally requiring time and manual effort. Thus, a challenging scenario arises when the target domain of application has no annotated dataset available, making tasks in such situation to lean on a training dataset of a different domain. Sharing this issue, object detection is a vital task for autonomous vehicles where the large amount of driving scenarios yields several domains of application requiring annotated data for the training process. In this work, a method for training a car detection system with annotated data from a source domain (day images) without requiring the image annotations of the target domain (night images) is presented. For that, a model based on Generative Adversarial Networks (GANs) is explored to enable the generation of an artificial dataset with its respective annotations. The artificial dataset (fake dataset) is created translating images from day-time domain to night-time domain. The fake dataset, which comprises annotated images of only the target domain (night images), is then used to train the car detector model. Experimental results showed that the proposed method achieved significant and consistent improvements, including the increasing by more than 10% of the detection performance when compared to the training with only the available annotated data (i.e., day images).

Deep learning has been successfully applied to several problems related to autonomous driving. Often, these solutions rely on large networks that require databases of real image samples of the problem (i.e., real world) for proper training. The acquisition of such real-world data sets is not always possible in the autonomous driving context, and sometimes their annotation is not feasible (e.g., takes too long or is too expensive). Moreover, in many tasks, there is an intrinsic data imbalance that most learning-based methods struggle to cope with. It turns out that traffic sign detection is a problem in which these three issues are seen altogether. In this work, we propose a novel database generation method that requires only (i) arbitrary natural images, i.e., requires no real image from the domain of interest, and (ii) templates of the traffic signs, i.e., templates synthetically created to illustrate the appearance of the category of a traffic sign. The effortlessly generated training database is shown to be effective for the training of a deep detector (such as Faster R-CNN) on German traffic signs, achieving 95.66% of mAP on average. In addition, the proposed method is able to detect traffic signs with an average precision, recall and F1-score of about 94%, 91% and 93%, respectively. The experiments surprisingly show that detectors can be trained with simple data generation methods and without problem domain data for the background, which is in the opposite direction of the common sense for deep learning.

Autonomous terrestrial vehicles must be capable of perceiving traffic lights and recognizing their current states to share the streets with human drivers. Most of the time, human drivers can easily identify the relevant traffic lights. To deal with this issue, a common solution for autonomous cars is to integrate recognition with prior maps. However, additional solution is required for the detection and recognition of the traffic light. Deep learning techniques have showed great performance and power of generalization including traffic related problems. Motivated by the advances in deep learning, some recent works leveraged some state-of-the-art deep detectors to locate (and further recognize) traffic lights from 2D camera images. However, none of them combine the power of the deep learning-based detectors with prior maps to recognize the state of the relevant traffic lights. Based on that, this work proposes to integrate the power of deep learning-based detection with the prior maps used by our car platform IARA (acronym for Intelligent Autonomous Robotic Automobile) to recognize the relevant traffic lights of predefined routes. The process is divided in two phases: an offline phase for map construction and traffic lights annotation; and an online phase for traffic light recognition and identification of the relevant ones. The proposed system was evaluated on five test cases (routes) in the city of Vit\'oria, each case being composed of a video sequence and a prior map with the relevant traffic lights for the route. Results showed that the proposed technique is able to correctly identify the relevant traffic light along the trajectory.

We present the Model-Predictive Motion Planner (MPMP) of the Intelligent Autonomous Robotic Automobile (IARA). IARA is a fully autonomous car that uses a path planner to compute a path from its current position to the desired destination. Using this path, the current position, a goal in the path and a map, IARA's MPMP is able to compute smooth trajectories from its current position to the goal in less than 50 ms. MPMP computes the poses of these trajectories so that they follow the path closely and, at the same time, are at a safe distance of eventual obstacles. Our experiments have shown that MPMP is able to compute trajectories that precisely follow a path produced by a Human driver (distance of 0.15 m in average) while smoothly driving IARA at speeds of up to 32.4 km/h (9 m/s).

The Brazilian government often misclassifies the goods it buys. That makes it hard to audit government expenditures. We cannot know whether the price paid for a ballpoint pen (code #7510) was reasonable if the pen was misclassified as a technical drawing pen (code #6675) or as any other good. This paper shows how we can use machine learning to reduce misclassification. I trained a support vector machine (SVM) classifier that takes a product description as input and returns the most likely category codes as output. I trained the classifier using 20 million goods purchased by the Brazilian government between 1999-04-01 and 2015-04-02. In 83.3% of the cases the correct category code was one of the three most likely category codes identified by the classifier. I used the trained classifier to develop a web app that might help the government reduce misclassification. I open sourced the code on GitHub; anyone can use and modify it.

This paper uses natural language processing to create the first machine-coded democracy index, which I call Automated Democracy Scores (ADS). The ADS are based on 42 million news articles from 6,043 different sources and cover all independent countries in the 1993-2012 period. Unlike the democracy indices we have today the ADS are replicable and have standard errors small enough to actually distinguish between cases. The ADS are produced with supervised learning. Three approaches are tried: a) a combination of Latent Semantic Analysis and tree-based regression methods; b) a combination of Latent Dirichlet Allocation and tree-based regression methods; and c) the Wordscores algorithm. The Wordscores algorithm outperforms the alternatives, so it is the one on which the ADS are based. There is a web application where anyone can change the training set and see how the results change: democracy-scores.org

Smartphones and portable devices have become ubiquitous and part of everyone's life. Due to the fact of its portability, these devices are perfect to record individuals' traces and life-logging generating vast amounts of data at low costs. These data is emerging as a new source for studies in human mobility patterns raising the number of research projects and techniques aiming to analyze and retrieve useful information from it. The aim of this paper is to explore GPS raw data from different individuals in a community and apply data mining algorithms to identify meaningful places in a region and describe user's profiles and its similarities. We evaluate the proposed method with a real-world dataset. The experimental results show that the steps performed to identify points of interest (POIs) and further the similarity between the users are quite satisfactory serving as a supplement for urban planning and social networks.

The algorithms of multi-objective optimisation had a relative growth in the last years. Thereby, it's requires some way of comparing the results of these. In this sense, performance measures play a key role. In general, it's considered some properties of these algorithms such as capacity, convergence, diversity or convergence-diversity. There are some known measures such as generational distance (GD), inverted generational distance (IGD), hypervolume (HV), Spread($\Delta$), Averaged Hausdorff distance ($\Delta_p$), R2-indicator, among others. In this paper, we focuses on proposing a new indicator to measure convergence based on the traditional formula for Shannon entropy. The main features about this measure are: 1) It does not require tho know the true Pareto set and 2) Medium computational cost when compared with Hypervolume.

One form of characterizing the expressiveness of a piecewise linear neural network is by the number of linear regions, or pieces, of the function modeled. We have observed substantial progress in this topic through lower and upper bounds on the maximum number of linear regions and a counting procedure. However, these bounds only account for the dimensions of the network and the exact counting may take a prohibitive amount of time, therefore making it infeasible to benchmark the expressiveness of networks. In this work, we approximate the number of linear regions of specific rectifier networks with an algorithm for probabilistic lower bounds of mixed-integer linear sets. In addition, we present a tighter upper bound that leverages network coefficients. We test both on trained networks. The algorithm for probabilistic lower bounds is several orders of magnitude faster than exact counting and the values reach similar orders of magnitude, hence making our approach a viable method to compare the expressiveness of such networks. The refined upper bound is particularly stronger on networks with narrow layers.

Referring expression generation (REG) models that use speaker-dependent information require a considerable amount of training data produced by every individual speaker, or may otherwise perform poorly. In this work we present a simple REG experiment that allows the use of larger training data sets by grouping speakers according to their overspecification preferences. Intrinsic evaluation shows that this method generally outperforms the personalised method found in previous work.

Complex networks refer to large-scale graphs with nontrivial connection patterns. The salient and interesting features that the complex network study offer in comparison to graph theory are the emphasis on the dynamical properties of the networks and the ability of inherently uncovering pattern formation of the vertices. In this paper, we present a hybrid data classification technique combining a low level and a high level classifier. The low level term can be equipped with any traditional classification techniques, which realize the classification task considering only physical features (e.g., geometrical or statistical features) of the input data. On the other hand, the high level term has the ability of detecting data patterns with semantic meanings. In this way, the classification is realized by means of the extraction of the underlying network's features constructed from the input data. As a result, the high level classification process measures the compliance of the test instances with the pattern formation of the training data. Out of various high level perspectives that can be utilized to capture semantic meaning, we utilize the dynamical features that are generated from a tourist walker in a networked environment. Specifically, a weighted combination of transient and cycle lengths generated by the tourist walk is employed for that end. Interestingly, our study shows that the proposed technique is able to further improve the already optimized performance of traditional classification techniques.

Deep neural networks have been successful in many predictive modeling tasks, such as image and language recognition, where large neural networks are often used to obtain good accuracy. Consequently, it is challenging to deploy these networks under limited computational resources, such as in mobile devices. In this work, we introduce an algorithm that removes units and layers of a neural network while not changing the output that is produced, which thus implies a lossless compression. This algorithm, which we denote as LEO (Lossless Expressiveness Optimization), relies on Mixed-Integer Linear Programming (MILP) to identify Rectifier Linear Units (ReLUs) with linear behavior over the input domain. By using L1 regularization to induce such behavior, we can benefit from training over a larger architecture than we would later use in the environment where the trained neural network is deployed.

Many aspects of life are associated with places of human mobility patterns and nowadays we are facing an increase in the pervasiveness of mobile devices these individuals carry. Positioning technologies that serve these devices such as the cellular antenna (GSM networks), global navigation satellite systems (GPS), and more recently the WiFi positioning system (WPS) provide large amounts of spatio-temporal data in a continuous way. Therefore, detecting significant places and the frequency of movements between them is fundamental to understand human behavior. In this paper, we propose a method for discovering user habits without any a priori or external knowledge by introducing a density-based clustering for spatio-temporal data to identify meaningful places and by applying a Gaussian Mixture Model (GMM) over the set of meaningful places to identify the representations of individual habits. To evaluate the proposed method we use two real-world datasets. One dataset contains high-density GPS data and the other one contains GSM mobile phone data in a coarse representation. The results show that the proposed method is suitable for this task as many unique habits were identified. This can be used for understanding users' behavior and to draw their characterizing profiles having a panorama of the mobility patterns from the data.

Two networks are equivalent if they produce the same output for any given input. In this paper, we study the possibility of transforming a deep neural network to another network with a different number of units or layers, which can be either equivalent, a local exact approximation, or a global linear approximation of the original network. On the practical side, we show that certain rectified linear units (ReLUs) can be safely removed from a network if they are always active or inactive for any valid input. If we only need an equivalent network for a smaller domain, then more units can be removed and some layers collapsed. On the theoretical side, we constructively show that for any feed-forward ReLU network, there exists a global linear approximation to a 2-hidden-layer shallow network with a fixed number of units. This result is a balance between the increasing number of units for arbitrary approximation with a single layer and the known upper bound of $\lceil log(n_0+1)\rceil +1$ layers for exact representation, where $n_0$ is the input dimension. While the transformed network may require an exponential number of units to capture the activation patterns of the original network, we show that it can be made substantially smaller by only accounting for the patterns that define linear regions. Based on experiments with ReLU networks on the MNIST dataset, we found that $l_1$-regularization and adversarial training reduces the number of linear regions significantly as the number of stable units increases due to weight sparsity. Therefore, we can also intentionally train ReLU networks to allow for effective loss-less compression and approximation.

We investigate the complexity of deep neural networks (DNN) that represent piecewise linear (PWL) functions. In particular, we study the number of linear regions, i.e. pieces, that a PWL function represented by a DNN can attain, both theoretically and empirically. We present (i) tighter upper and lower bounds for the maximum number of linear regions on rectifier networks, which are exact for inputs of dimension one; (ii) a first upper bound for multi-layer maxout networks; and (iii) a first method to perform exact enumeration or counting of the number of regions by modeling the DNN with a mixed-integer linear formulation. These bounds come from leveraging the dimension of the space defining each linear region. The results also indicate that a deep rectifier network can only have more linear regions than every shallow counterpart with same number of neurons if that number exceeds the dimension of the input.