Exact Combinatorial Optimization with Graph Convolutional Neural Networks

Combinatorial optimization problems are typically tackled by the branch-and-bound paradigm. We propose to learn a variable selection policy for branch-and-bound in mixed-integer linear programming, by imitation learning on a diversified variant of the strong branching expert rule. We encode states as bipartite graphs and parameterize the policy as a graph convolutional neural network. Experiments on a series of synthetic problems demonstrate that our approach produces policies that can improve upon expert-designed branching rules on large problems, and generalize to instances significantly larger than seen during training.

Optimizing AI at the Edge: from network topology design to MCU deployment

The first topic analyzed in the thesis will be Neural Architecture Search (NAS). I will focus on two different tools that I developed, one to optimize the architecture of Temporal Convolutional Networks (TCNs), a convolutional model for time-series processing that has recently emerged, and one to optimize the data precision of tensors inside CNNs. The first NAS proposed explicitly targets the optimization of the most peculiar architectural parameters of TCNs, namely dilation, receptive field, and the number of features in each layer. Note that this is the first NAS that explicitly targets these networks. The second NAS proposed instead focuses on finding the most efficient data format for a target CNN, with the granularity of the layer filter. Note that applying these two NASes in sequence allows an "application designer" to minimize the structure of the neural network employed, minimizing the number of operations or the memory usage of the network. After that, the second topic described is the optimization of neural network deployment on edge devices. Importantly, exploiting edge platforms' scarce resources is critical for NN efficient execution on MCUs. To do so, I will introduce DORY (Deployment Oriented to memoRY) -- an automatic tool to deploy CNNs on low-cost MCUs. DORY, in different steps, can manage different levels of memory inside the MCU automatically, offload the computation workload (i.e., the different layers of a neural network) to dedicated hardware accelerators, and automatically generates ANSI C code that orchestrates off- and on-chip transfers with the computation phases. On top of this, I will introduce two optimized computation libraries that DORY can exploit to deploy TCNs and Transformers on edge efficiently. I conclude the thesis with two different applications on bio-signal analysis, i.e., heart rate tracking and sEMG-based gesture recognition.

Predicting corticospinal excitability from pre-TMS EEG using convolutional neural networks

The amplitude of motor evoked potentials (MEPs) elicited by transcranial magnetic stimulation (TMS) of the primary motor cortex (M1) shows a large variability from trial to trial, although MEPs are evoked by the same repeated stimulus. A multitude of factors is believed to influence MEP amplitudes, such as cortical, spinal and motor excitability state. The goal of this work is to explore to which degree the variation in MEP amplitudes can be explained by the cortical state right before the stimulation. Specifically, we analyzed a dataset acquired on eleven healthy subjects comprising, for each subject, 840 single TMS pulses applied to the left M1 during acquisition of electroencephalography (EEG) and electromyography (EMG). An interpretable convolutional neural network, named SincEEGNet, was utilized to discriminate between low- and high-corticospinal excitability trials, defined according to the MEP amplitude, using in input the pre-TMS EEG. This data-driven approach enabled considering multiple brain locations and frequency bands without any a priori selection. Post-hoc interpretation techniques were adopted to enhance interpretation by identifying the more relevant EEG features for the classification. Results show that individualized classifiers successfully discriminated between low and high M1 excitability states in all participants. Outcomes of the interpretation methods suggest the importance of the electrodes situated over the TMS stimulation site, as well as the relevance of the temporal samples of the input EEG closer to the stimulation time. This novel decoding method allows causal investigation of the cortical excitability state, which may be relevant for personalizing and increasing the efficacy of therapeutic brain-state dependent brain stimulation (for example in patients affected by Parkinson’s disease).

Mitigating non-Lambertian surfaces issues in Stereo Matching with Neural Radiance Fields

Depth estimation from images has long been regarded as a preferable alternative compared to expensive and intrusive active sensors, such as LiDAR and ToF. The topic has attracted the attention of an increasingly wide audience thanks to the great amount of application domains, such as autonomous driving, robotic navigation and 3D reconstruction. Among the various techniques employed for depth estimation, stereo matching is one of the most widespread, owing to its robustness, speed and simplicity in setup. Recent developments has been aided by the abundance of annotated stereo images, which granted to deep learning the opportunity to thrive in a research area where deep networks can reach state-of-the-art sub-pixel precision in most cases. Despite the recent findings, stereo matching still begets many open challenges, two among them being finding pixel correspondences in presence of objects that exhibits a non-Lambertian behaviour and processing high-resolution images. Recently, a novel dataset named Booster, which contains high-resolution stereo pairs featuring a large collection of labeled non-Lambertian objects, has been released. The work shown that training state-of-the-art deep neural network on such data improves the generalization capabilities of these networks also in presence of non-Lambertian surfaces. Regardless being a further step to tackle the aforementioned challenge, Booster includes a rather small number of annotated images, and thus cannot satisfy the intensive training requirements of deep learning. This thesis work aims to investigate novel view synthesis techniques to augment the Booster dataset, with ultimate goal of improving stereo matching reliability in presence of high-resolution images that displays non-Lambertian surfaces.

Tree regularized neural networks for key point analysis

This thesis contributes to the ArgMining 2021 shared task on Key Point Analysis. Key Point Analysis entails extracting and calculating the prevalence of a concise list of the most prominent talking points, from an input corpus. These talking points are usually referred to as key points. Key point analysis is divided into two subtasks: Key Point Matching, which involves assigning a matching score to each key point/argument pair, and Key Point Generation, which consists of the generation of key points. The task of Key Point Matching was approached using different models: a pretrained Sentence Transformers model and a tree-constrained Graph Neural Network were tested. The best model was the fine-tuned Sentence Transformers, which achieved a mean Average Precision score of 0.75, ranking 12 compared to other participating teams. The model was then used for the subtask of Key Point Generation using the extractive method in the selection of key point candidates and the model developed for the previous subtask to evaluate them.

Neural Radiance Fields for Relighting and View Synthesis

Neural scene representation and neural rendering are new computer vision techniques that enable the reconstruction and implicit representation of real 3D scenes from a set of 2D captured images, by fitting a deep neural network. The trained network can then be used to render novel views of the scene. A recent work in this field, Neural Radiance Fields (NeRF), presented a state-of-the-art approach, which uses a simple Multilayer Perceptron (MLP) to generate photo-realistic RGB images of a scene from arbitrary viewpoints. However, NeRF does not model any light interaction with the fitted scene; therefore, despite producing compelling results for the view synthesis task, it does not provide a solution for relighting. In this work, we propose a new architecture to enable relighting capabilities in NeRF-based representations and we introduce a new real-world dataset to train and evaluate such a model. Our method demonstrates the ability to perform realistic rendering of novel views under arbitrary lighting conditions.

Explosive Synchronization With Partial Degree-frequency Correlation.

Networks of Kuramoto oscillators with a positive correlation between the oscillators frequencies and the degree of their corresponding vertices exhibit so-called explosive synchronization behavior, which is now under intensive investigation. Here we study and discuss explosive synchronization in a situation that has not yet been considered, namely when only a part, typically a small part, of the vertices is subjected to a degree-frequency correlation. Our results show that in order to have explosive synchronization, it suffices to have degree-frequency correlations only for the hubs, the vertices with the highest degrees. Moreover, we show that a partial degree-frequency correlation does not only promotes but also allows explosive synchronization to happen in networks for which a full degree-frequency correlation would not allow it. We perform a mean-field analysis and our conclusions were corroborated by exhaustive numerical experiments for synthetic networks and also for the undirected and unweighed version of a typical benchmark biological network, namely the neural network of the worm Caenorhabditis elegans. The latter is an explicit example where partial degree-frequency correlation leads to explosive synchronization with hysteresis, in contrast with the fully correlated case, for which no explosive synchronization is observed.

Uso de redes neurais artificiais na identificação de vocalização de suínos

Animal welfare has been an important research topic in animal production mainly in its ways of assessment. Vocalization is found to be an interesting tool for evaluating welfare as it provides data in a non-invasive way as well as it allows easy automation of process. The present research had as objective the implementation of an algorithm based on artificial neural network that had the potential of identifying vocalization related to welfare pattern indicatives. The research was done in two parts, the first was the development of the algorithm, and the second its validation with data from the field. Previous records allowed the development of the algorithm from behaviors observed in sows housed in farrowing cages. Matlab® software was used for implementing the network. It was selected a retropropagation gradient algorithm for training the network with the following stop criteria: maximum of 5,000 interactions or error quadratic addition smaller than 0.1. Validation was done with sows and piglets housed in commercial farm. Among the usual behaviors the ones that deserved enhancement were: the feed dispute at farrowing and the eventual risk of involuntary aggression between the piglets or between those and the sow. The algorithm was able to identify through the noise intensity the inherent risk situation of piglets welfare reduction.

Inteligência artificial aplicada à Zootecnia

Os sistemas biológicos são surpreendentemente flexíveis pra processar informação proveniente do mundo real. Alguns organismos biológicos possuem uma unidade central de processamento denominada de cérebro. O cérebro humano consiste de 10(11) neurônios e realiza processamento inteligente de forma exata e subjetiva. A Inteligência Artificial (IA) tenta trazer para o mundo da computação digital a heurística dos sistemas biológicos de várias maneiras, mas, ainda resta muito para que isso seja concretizado. No entanto, algumas técnicas como Redes neurais artificiais e lógica fuzzy tem mostrado efetivas para resolver problemas complexos usando a heurística dos sistemas biológicos. Recentemente o numero de aplicação dos métodos da IA em sistemas zootécnicos tem aumentado significativamente. O objetivo deste artigo é explicar os princípios básicos da resolução de problemas usando heurística e demonstrar como a IA pode ser aplicada para construir um sistema especialista para resolver problemas na área de zootecnia.

Multidisciplinary optimization of a catamaran's hydrofoil design with dynamic response surface and distributive processing

This paper presents a rational approach to the design of a catamaran's hydrofoil applied within a modern context of multidisciplinary optimization. The approach used includes the use of response surfaces represented by neural networks and a distributed programming environment that increases the optimization speed. A rational approach to the problem simplifies the complex optimization model; when combined with the distributed dynamic training used for the response surfaces, this model increases the efficiency of the process. The results achieved using this approach have justified this publication.

Observational Characterization of the Downward Atmospheric Longwave Radiation at the Surface in the City of Sao Paulo

This work describes the seasonal and diurnal variations of downward longwave atmospheric irradiance (LW) at the surface in Sao Paulo, Brazil, using 5-min-averaged values of LW, air temperature, relative humidity, and solar radiation observed continuously and simultaneously from 1997 to 2006 on a micrometeorological platform, located at the top of a 4-story building. An objective procedure, including 2-step filtering and dome emission effect correction, was used to evaluate the quality of the 9-yr-long LW dataset. The comparison between LW values observed and yielded by the Surface Radiation Budget project shows spatial and temporal agreement, indicating that monthly and annual average values of LW observed in one point of Sao Paulo can be used as representative of the entire metropolitan region of Sao Paulo. The maximum monthly averaged value of the LW is observed during summer (389 +/- 14 W m(-2): January), and the minimum is observed during winter (332 +/- 12 W m(-2); July). The effective emissivity follows the LW and shows a maximum in summer (0.907 +/- 0.032; January) and a minimum in winter (0.818 +/- 0.029; June). The mean cloud effect, identified objectively by comparing the monthly averaged values of the LW during clear-sky days and all-sky conditions, intensified the monthly average LW by about 32.0 +/- 3.5 W m(-2) and the atmospheric effective emissivity by about 0.088 +/- 0.024. In August, the driest month of the year in Sao Paulo, the diurnal evolution of the LW shows a minimum (325 +/- 11 W m(-2)) at 0900 LT and a maximum (345 12 W m-2) at 1800 LT, which lags behind (by 4 h) the maximum diurnal variation of the screen temperature. The diurnal evolution of effective emissivity shows a minimum (0.781 +/- 0.027) during daytime and a maximum (0.842 +/- 0.030) during nighttime. The diurnal evolution of all-sky condition and clear-sky day differences in the effective emissivity remain relatively constant (7% +/- 1%), indicating that clouds do not change the emissivity diurnal pattern. The relationship between effective emissivity and screen air temperature and between effective emissivity and water vapor is complex. During the night, when the planetary boundary layer is shallower, the effective emissivity can be estimated by screen parameters. During the day, the relationship between effective emissivity and screen parameters varies from place to place and depends on the planetary boundary layer process. Because the empirical expressions do not contain enough information about the diurnal variation of the vertical stratification of air temperature and moisture in Sao Paulo, they are likely to fail in reproducing the diurnal variation of the surface emissivity. The most accurate way to estimate the LW for clear-sky conditions in Sao Paulo is to use an expression derived from a purely empirical approach.

Social interaction as a heuristic for combinatorial optimization problems

We investigate the performance of a variant of Axelrod's model for dissemination of culture-the Adaptive Culture Heuristic (ACH)-on solving an NP-Complete optimization problem, namely, the classification of binary input patterns of size F by a Boolean Binary Perceptron. In this heuristic, N agents, characterized by binary strings of length F which represent possible solutions to the optimization problem, are fixed at the sites of a square lattice and interact with their nearest neighbors only. The interactions are such that the agents' strings (or cultures) become more similar to the low-cost strings of their neighbors resulting in the dissemination of these strings across the lattice. Eventually the dynamics freezes into a homogeneous absorbing configuration in which all agents exhibit identical solutions to the optimization problem. We find through extensive simulations that the probability of finding the optimal solution is a function of the reduced variable F/N(1/4) so that the number of agents must increase with the fourth power of the problem size, N proportional to F(4), to guarantee a fixed probability of success. In this case, we find that the relaxation time to reach an absorbing configuration scales with F(6) which can be interpreted as the overall computational cost of the ACH to find an optimal set of weights for a Boolean binary perceptron, given a fixed probability of success.

Simultaneous optimization by neuro-genetic approach for analysis of plant materials by laser induced breakdown spectroscopy

A simultaneous optimization strategy based on a neuro-genetic approach is proposed for selection of laser induced breakdown spectroscopy operational conditions for the simultaneous determination of macronutrients (Ca, Mg and P), micro-nutrients (B, Cu, Fe, Mn and Zn), Al and Si in plant samples. A laser induced breakdown spectroscopy system equipped with a 10 Hz Q-switched Nd:YAG laser (12 ns, 532 nm, 140 mJ) and an Echelle spectrometer with intensified coupled-charge device was used. Integration time gate, delay time, amplification gain and number of pulses were optimized. Pellets of spinach leaves (NIST 1570a) were employed as laboratory samples. In order to find a model that could correlate laser induced breakdown spectroscopy operational conditions with compromised high peak areas of all elements simultaneously, a Bayesian Regularized Artificial Neural Network approach was employed. Subsequently, a genetic algorithm was applied to find optimal conditions for the neural network model, in an approach called neuro-genetic, A single laser induced breakdown spectroscopy working condition that maximizes peak areas of all elements simultaneously, was obtained with the following optimized parameters: 9.0 mu s integration time gate, 1.1 mu s delay time, 225 (a.u.) amplification gain and 30 accumulated laser pulses. The proposed approach is a useful and a suitable tool for the optimization process of such a complex analytical problem. (C) 2009 Elsevier B.V. All rights reserved.

Kernel machines for epilepsy diagnosis via EEG signal classification: A comparative study

Objective: We carry out a systematic assessment on a suite of kernel-based learning machines while coping with the task of epilepsy diagnosis through automatic electroencephalogram (EEG) signal classification. Methods and materials: The kernel machines investigated include the standard support vector machine (SVM), the least squares SVM, the Lagrangian SVM, the smooth SVM, the proximal SVM, and the relevance vector machine. An extensive series of experiments was conducted on publicly available data, whose clinical EEG recordings were obtained from five normal subjects and five epileptic patients. The performance levels delivered by the different kernel machines are contrasted in terms of the criteria of predictive accuracy, sensitivity to the kernel function/parameter value, and sensitivity to the type of features extracted from the signal. For this purpose, 26 values for the kernel parameter (radius) of two well-known kernel functions (namely. Gaussian and exponential radial basis functions) were considered as well as 21 types of features extracted from the EEG signal, including statistical values derived from the discrete wavelet transform, Lyapunov exponents, and combinations thereof. Results: We first quantitatively assess the impact of the choice of the wavelet basis on the quality of the features extracted. Four wavelet basis functions were considered in this study. Then, we provide the average accuracy (i.e., cross-validation error) values delivered by 252 kernel machine configurations; in particular, 40%/35% of the best-calibrated models of the standard and least squares SVMs reached 100% accuracy rate for the two kernel functions considered. Moreover, we show the sensitivity profiles exhibited by a large sample of the configurations whereby one can visually inspect their levels of sensitiveness to the type of feature and to the kernel function/parameter value. Conclusions: Overall, the results evidence that all kernel machines are competitive in terms of accuracy, with the standard and least squares SVMs prevailing more consistently. Moreover, the choice of the kernel function and parameter value as well as the choice of the feature extractor are critical decisions to be taken, albeit the choice of the wavelet family seems not to be so relevant. Also, the statistical values calculated over the Lyapunov exponents were good sources of signal representation, but not as informative as their wavelet counterparts. Finally, a typical sensitivity profile has emerged among all types of machines, involving some regions of stability separated by zones of sharp variation, with some kernel parameter values clearly associated with better accuracy rates (zones of optimality). (C) 2011 Elsevier B.V. All rights reserved.

Identification of the biodiesel source using an electronic nose

Biodiesel is an important new alternative fuel. The feedstock used and the process employed determines whether it fulfills the required specifications. In this work, an identification method is proposed using an electronic nose (e-nose). Four samples of biodiesel from different sources and one of petrodiesel were analyzed and well-recognized by the e-nose. Both pure biodiesel and B20 blends were studied. Furthermore, an innovative semiquantitative method is proposed on the basis of the smellprints correlated by a feed-forward artificial neural network. The results have demonstrated that the e-nose can be used to identify the biodiesel source and as a preliminary quantitative assay in place of expensive equipment.