Flexible Computing Systems For AI Acceleration At The Extreme Edge Of The IoT

Embedding intelligence in extreme edge devices allows distilling raw data acquired from sensors into actionable information, directly on IoT end-nodes. This computing paradigm, in which end-nodes no longer depend entirely on the Cloud, offers undeniable benefits, driving a large research area (TinyML) to deploy leading Machine Learning (ML) algorithms on micro-controller class of devices. To fit the limited memory storage capability of these tiny platforms, full-precision Deep Neural Networks (DNNs) are compressed by representing their data down to byte and sub-byte formats, in the integer domain. However, the current generation of micro-controller systems can barely cope with the computing requirements of QNNs. This thesis tackles the challenge from many perspectives, presenting solutions both at software and hardware levels, exploiting parallelism, heterogeneity and software programmability to guarantee high flexibility and high energy-performance proportionality. The first contribution, PULP-NN, is an optimized software computing library for QNN inference on parallel ultra-low-power (PULP) clusters of RISC-V processors, showing one order of magnitude improvements in performance and energy efficiency, compared to current State-of-the-Art (SoA) STM32 micro-controller systems (MCUs) based on ARM Cortex-M cores. The second contribution is XpulpNN, a set of RISC-V domain specific instruction set architecture (ISA) extensions to deal with sub-byte integer arithmetic computation. The solution, including the ISA extensions and the micro-architecture to support them, achieves energy efficiency comparable with dedicated DNN accelerators and surpasses the efficiency of SoA ARM Cortex-M based MCUs, such as the low-end STM32M4 and the high-end STM32H7 devices, by up to three orders of magnitude. To overcome the Von Neumann bottleneck while guaranteeing the highest flexibility, the final contribution integrates an Analog In-Memory Computing accelerator into the PULP cluster, creating a fully programmable heterogeneous fabric that demonstrates end-to-end inference capabilities of SoA MobileNetV2 models, showing two orders of magnitude performance improvements over current SoA analog/digital solutions.

Edge and Big Data technologies for Industry 4.0 to create an integrated pre-sale and after-sale environment

The fourth industrial revolution, also known as Industry 4.0, has rapidly gained traction in businesses across Europe and the world, becoming a central theme in small, medium, and large enterprises alike. This new paradigm shifts the focus from locally-based and barely automated firms to a globally interconnected industrial sector, stimulating economic growth and productivity, and supporting the upskilling and reskilling of employees. However, despite the maturity and scalability of information and cloud technologies, the support systems already present in the machine field are often outdated and lack the necessary security, access control, and advanced communication capabilities. This dissertation proposes architectures and technologies designed to bridge the gap between Operational and Information Technology, in a manner that is non-disruptive, efficient, and scalable. The proposal presents cloud-enabled data-gathering architectures that make use of the newest IT and networking technologies to achieve the desired quality of service and non-functional properties. By harnessing industrial and business data, processes can be optimized even before product sale, while the integrated environment enhances data exchange for post-sale support. The architectures have been tested and have shown encouraging performance results, providing a promising solution for companies looking to embrace Industry 4.0, enhance their operational capabilities, and prepare themselves for the upcoming fifth human-centric revolution.

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.

Development and characterization of cutting-edge equipment for EMC measurements and advanced power line filter design

This Doctoral Thesis aims at studying, developing, and characterizing cutting edge equipment for EMC measurements and proposing innovative and advanced power line filter design techniques. This document summarizes a three-year work, is strictly industry oriented and relies on EMC standards and regulations. It contains the main results, findings, and effort with the purpose of bringing innovative contributions at the scientific community. Conducted emissions interferences are usually suppressed with power line filters. These filters are composed by common mode chokes, X capacitors and Y capacitors in order to mitigate both the differential mode and common mode noise, which compose the overall conducted emissions. However, even at present days, available power line filter design techniques show several disadvantages. First of all, filters are designed to be implemented in ideal 50 Ω systems, condition which is far away from reality. Then, the attenuation introduced by the filter for common or differential mode noise is analyzed independently, without considering the possible mode conversion that can be produced by impedance mismatches, or asymmetries in either the power line filter itself or the equipment under test. Ultimately, the instrumentation used to perform conducted emissions measurement is, in most cases, not adequate. All these factors lead to an inaccurate design, contributing at increasing the size of the filter, making it more expensive and less performant than it should be.

Electronic And Structural Study Of Pt-modified Au Vicinal Surfaces: A Model System For Pt-au Catalysts.

Two single crystalline surfaces of Au vicinal to the (111) plane were modified with Pt and studied using scanning tunneling microscopy (STM) and X-ray photoemission spectroscopy (XPS) in ultra-high vacuum environment. The vicinal surfaces studied are Au(332) and Au(887) and different Pt coverage (θPt) were deposited on each surface. From STM images we determine that Pt deposits on both surfaces as nanoislands with heights ranging from 1 ML to 3 ML depending on θPt. On both surfaces the early growth of Pt ad-islands occurs at the lower part of the step edge, with Pt ad-atoms being incorporated into the steps in some cases. XPS results indicate that partial alloying of Pt occurs at the interface at room temperature and at all coverage, as suggested by the negative chemical shift of Pt 4f core line, indicating an upward shift of the d-band center of the alloyed Pt. Also, the existence of a segregated Pt phase especially at higher coverage is detected by XPS. Sample annealing indicates that the temperature rise promotes a further incorporation of Pt atoms into the Au substrate as supported by STM and XPS results. Additionally, the catalytic activity of different PtAu systems reported in the literature for some electrochemical reactions is discussed considering our findings.

Mechanical Properties And Fracture Dynamics Of Silicene Membranes.

As graphene has become one of the most important materials, there is renewed interest in other similar structures. One example is silicene, the silicon analogue of graphene. It shares some of the remarkable graphene properties, such as the Dirac cone, but presents some distinct ones, such as a pronounced structural buckling. We have investigated, through density functional based tight-binding (DFTB), as well as reactive molecular dynamics (using ReaxFF), the mechanical properties of suspended single-layer silicene. We calculated the elastic constants, analyzed the fracture patterns and edge reconstructions. We also addressed the stress distributions, unbuckling mechanisms and the fracture dependence on the temperature. We analysed the differences due to distinct edge morphologies, namely zigzag and armchair.

Tree structure and richness in an Atlantic Forest fragment: distance from anthropogenic and natural edges

Approximately 7.2% of the Atlantic rainforest remains in Brazil, with only 16% of this forest remaining in the State of Rio de Janeiro, all of it distributed in fragments. This forest fragmentation can produce biotic and abiotic differences between edges and the fragment interior. In this study, we compared the structure and richness of tree communities in three habitats - an anthropogenic edge (AE), a natural edge (NE) and the fragment interior (FI) - of a fragment of Atlantic forest in the State of Rio de Janeiro, Brazil (22°50'S and 42°28'W). One thousand and seventy-six trees with a diameter at breast height > 4.8 cm, belonging to 132 morphospecies and 39 families, were sampled in a total study area of 0.75 ha. NE had the greatest basal area and the trees in this habitat had the greatest diameter:height allometric coefficient, whereas AE had a lower richness and greater variation in the height of the first tree branch. Tree density, diameter, height and the proportion of standing dead trees did not differ among the habitats. There was marked heterogeneity among replicates within each habitat. These results indicate that the forest interior and the fragment edges (natural or anthropogenic) do not differ markedly considering the studied parameters. Other factors, such as the age from the edge, type of matrix and proximity of gaps, may play a more important role in plant community structure than the proximity from edges.

Metodologia para a suavização de dados biomecanicos por função não parametrica ponderada local robusta

Universidade Estadual de Campinas. Faculdade de Educação Física

Early treatment of Class III malocclusion: 10-year clinical follow-up

Angle Class III malocclusion has been a challenge for researchers concerning diagnosis, prognosis and treatment. It has a prevalence of 5% in the Brazilian population, and may have a genetic or environmental etiology. This malocclusion can be classified as dentoalveolar, skeletal or functional, which will determine the prognosis. Considering these topics, the aim of this study was to describe and discuss a clinical case with functional Class III malocclusion treated by a two-stage approach (interceptive and corrective), with a long-term follow-up. In this case, the patient was treated with a chincup and an Eschler arch, used simultaneously during 14 months, followed by corrective orthodontics. It should be noticed that, in this case, initial diagnosis at the centric relation allowed visualizing the anterior teeth in an edge-to-edge relationship, thereby favoring the prognosis. After completion of the treatment, the patient was followed for a 10-year period, and stability was observed. The clinical treatment results showed that it is possible to achieve favorable outcomes with early management in functional Class III malocclusion patients.

Micromorphological and hardness analyses of human and bovine sclerotic dentin: a comparative study

The purpose of this study was to test the hypothesis that both human and bovine sclerotic dentin have similar hardness properties, in addition to similar micromorphological characteristics. Sixteen teeth (8 human and 8 bovine) exhibiting exposed dentin in the incisal edge and showing characteristics typical of sclerosis were used. Vickers surface microhardness testing was conducted. Three areas of the dentin surface of each specimen were selected. All teeth were processed for scanning electron microscopy in order to estimate the amount (in percentage) of solid dentin on the sclerotic dentin surface. The data were compared by Student's t test (α = 0.05). The micromorphological and microhardness data were compared by Pearson's linear correlation test (α = 0.05). The mean percentages of solid dentin of human and bovine sclerotic dentin were similar (human 90.71 ± 0.83 and bovine 89.08 ± 0.81, p = 0.18). The mean microhardness value (VHN) of human sclerotic dentin was significantly higher than that of bovine sclerotic dentin (human 45.26 ± 2.92 and bovine 29.93 ± 3.83, p = 0.006). No correlation was found between the microhardness values and the amount of solid dentin in the sclerotic dentin, irrespective of the species considered (human R² = 0.0240, p = 0.714; bovine R² = 0.0017, p = 0.923; and combined R² = 0.038, p = 0.46). We concluded that although both bovine and human sclerotic dentin present a similar amount of solid tissue, human sclerotic dentin presents higher microhardness than bovine sclerotic dentin.

Mudanças no Código Florestal e seu impacto na ecologia e diversidade dos mamíferos no Brasil

Os ecossistemas florestais do Brasil abrigam um dos mais altos níveis de diversidade de mamíferos da Terra, e boa parte dessa diversidade se encontra nas áreas legalmente protegidas em áreas de domínio privado. As reservas legais (RLs) e áreas de proteção permanente (APPs) representam estratégias importantes para a proteção e manutenção dessa diversidade. Mudanças propostas no Código Florestal certamente trarão efeitos irreversíveis para a diversidade de mamíferos no Brasil. Os mamíferos apresentam papéis-chave nos ecossistemas, atuando como polinizadores e dispersores de sementes. A extinção local de algumas espécies pode reduzir os serviços ecológicos nas RLs e APPs. Outra consequência grave da redução de áreas de vegetação nativa caso a mudança no Código Florestal seja aprovada será o aumento no risco de transmição de doenças, trazendo sério problemas a saúde pública no Brasil.

Diversidade, distribuição e abundância de flebotomíneos (Diptera: Psychodidae) no Paraná

Devido à crescente expansão da leishmaniose visceral americana (LVA) no Brasil, o presente estudo teve como objetivo identificar as espécies de flebotomíneos em áreas vulneráveis à transmissão dessa parasitose, bem como em outras sem qualquer informação sobre a presença desses dípteros no Paraná. As coletas de flebotomíneos foram realizadas em 46 localidades distribuídas em 37 municípios do Paraná, no período de março de 2004 a novembro de 2005. Em cada uma das localidades foram instaladas armadilhas de Falcão, durante três noites consecutivas, em vegetação natural e ambientes antrópicos (intra e peridomicílio). Ocasionalmente, foram instalas armadilhas de Shannon e feitas inspeções de paredes e aspiração em domicílio, peridomicílio e extradomicílio. O tratamento dos dados baseou-se na estimativa das freqüências e abundância das espécies, segundo cinco regiões de distintas paisagens originais. Coletaram-se 38.662 flebotomíneos de 23 espécies. Predominaram Nyssomyia neivai (Pinto) (75.6%), Ny. whitmani (Antunes & Coutinho) (10.1%), Migonemyia migonei (França) (7.8%), Expapillata firmatoi (Barreto et al.) (2.1%) and Pintomyia fischeri (Pinto) (1,6%); representando juntas 97,2% dos flebotomíneos coletados. Lutzomyia longipalpis (Lutz & Neiva) o principal vetor da LVA não foi encontrado. No entanto, capturou-se Lu. gaminarai (Cordero et al.), cujas fêmeas são morfologicamente semelhantes às de Lu. longipalpis. As espécies mais freqüentes e abundantes têm sido apontadas como vetores da leishmaniose tegumentar no Paraná e em outras áreas das Regiões Sudeste e Sul do Brasil. A presença de Lu. gaminarai no Paraná suscita a necessidade de estudos do seu comportamento, inclusive em relação à sua competência vetorial do agente da leishmaniose visceral.

Ultra-sonografia do fígado, aparelho renal e reprodutivo da jibóia (Boa constrictor)

Atualmente os animais silvestres têm despertado o interesse particular na criação domestica. Na medicina de animais selvagens, os exames ultra-sonográficos podem ser considerados como ferramenta para diagnosticar e prevenir doenças. Deste modo, realizou-se um estudo em 20 jibóias (Boa constrictor), a fim de caracterizar a morfologia e aparência ultra-sonográfica das estruturas presentes da cavidade celomática desses animais. Ultra-sonograficamente, o fígado apresentou-se variando de hipoecóica a levemente hiperecogênica, com margens ecogênicas e ecotextura homogênea em toda sua extensão. Os rins mostraram formato elipsóide, com cápsula fina, regular e hiperecóica. Os folículos ovarianos apresentaram formato ovóide, margens finas, regulares e discretamente hiperecóicas. As estruturas do sistema reprodutor do macho não foram evidenciadas com precisão, devido a sua ecogenicidade similar em relação às estruturas adjacentes e pela presença do "corpo gorduroso" localizado nessa região. A ultra-sonografia da cavidade celomática em jibóias demonstrou ser uma técnica rápida e de fácil acesso, permitindo identificar a morfologia, sintopia e aparência ultra-sonográfica de estruturas como o fígado, rins e de folículos vitelogênicos nas fêmeas.

Análise proliferativa nas células trofoblásticas em bovinos

The aim of the present study was to evaluate the cell proliferative activity, by AgNORs number, in different regions of bovine placenta throughout gestation. A total of 28 bovine placentas were separated into four groups: group I (60 to 120 days), group II (121 to 170 days), group III (171 to 220 days), and group IV (221 to 290 days). It was found a greater number of AgNORs in giant trophoblastic cells (GTC) when compared with mononuclear trophoblastic cells (MTC) (p<0,001) in all regions and gestational groups analyzed, that confirms their intensive synthesis activity in trophoblast epithelium. The central region of the placentome begins an intense proliferative activity in group II, observed by clusters, while placentomes edges showed a higher number of clusters on group III. These data suggest that the central region of the placentomes began an intense proliferative activity prior to its edge, both declines at the end of pregnancy. Interplacentomal area showed a higher number of AgNORs in the group IV, suggesting a higher proliferative activity of these cells at the end of pregnancy. The results of this study indicate that the proliferative activity, as determined by the amount of intranuclear AgNORs, exhibits patterns that are not only specific to each type of trophoblastic cells, but also for each specific region of bovine placenta throughout pregnancy.

Modelo de suprimento sanguíneo do intestino grosso do tamanduá bandeira (Myrmecophaga tridactyla)

Em seis animais adultos Myrmecophaga tridactyla estudou-se o modelo de suprimento sanguíneo do intestino grosso, que é dependente das artérias mesentérica cranial (AMCr) e caudal (AMC). Os espécimes coletados conforme as normas do IBAMA (Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis) foram perfundidos com água (40 C), injetados com látex corado, fixados em formol e conservados em solução alcoólica (50%). O mesocólon está disposto na linha sagital mediana e fixa o intestino grosso à parede dorsal do abdome. Derivaram da AMC: a artéria retal cranial, sete a 14 artérias cólicas e uma ou duas artérias ileocólicas, que apresentaram anastomoses de irrigação com a AMCr. A AMC finaliza-se na borda mesocólica das alças intestinais, emitindo ramos cólicos retos a partir das arcadas justacólicas, que penetram na intimidade da musculatura longitudinal. Ao longo do trajeto da AMC foram observadas ilhas arteriais, e a região ileocólica apresentou maior densidade vascular.