Search

Computer-aided diagnostic has become a thriving research area in recent years, namely on the identification of skin lesions such as melanoma. This work presents a novel approach to this field by exploiting the 3D characteristics of the skin lesion surface, advancing beyond common features such as, shape, colour, and texture, extracted from dermoscopic RGB images. To this end, a relevant set of features was investigated to obtain 3D skin lesion characteristics from images with depth information. These features were used to train a Bag-of-Features model to distinguish between malignant and benign lesions, also discriminating melanoma from all other lesion types. Despite the large class imbalance, often present in medical image datasets, the feature set achieved a top accuracy of 73.08%, comprising 75.00% sensitivity and 66.67% specificity when classifying between malignant and benign lesions, and 88.46% accuracy (100.00% sensitivity and 86.96% specificity) when discriminating melanoma from all other lesion images, using only depth information. The achieved experimental results indicate the existence of relevant discriminative characteristics in the 3D surface of skin lesions which allow the improvement of existing classification methods based on 2D image characteristics only.

Forecasting asset volatility is a common and important application in finance, often used as a measure of the future risk of an investment. With high-frequency data availability in the last few decades, new volatility models have been proposed that make use of realized volatility measures. These models have been empirically shown to provide better performance when compared with classical low-frequency approaches. In this paper, we propose a methodology combining an MLP neural network with heterogeneous autoregressive models for daily forecasting of the covariance matrix of a portfolio and for dynamically allocating optimal weights. Using Cholesky decomposition we guarantee the positive definite estimates. We used real historical data of 5 stocks of the Brazilian market in our experiments and the results demonstrated statistically superior forecasts compared with the baselines HAR, HARQ, ARFIMA, and DCC, using the Model Confidence Set test. We also evaluate the performance of the portfolios using the matrix estimates from the different models as parameters to Global Minimum Variance Portfolio. Our approach achieved the smallest values of portfolio variance compared with the baselines.

This work presents an integration of the Cop-peliaSim robotics simulator together with ROS in a scene inspired by a stage in the practical modality of the Brazilian Robotics Olympics (OBR). A differential-type robot with 5 proximity sensors and 6 luminance sensors was used in the project. The dimensional characteristics of the robot and the position of the sensors, in addition to navigation control and obstacle avoidance strategies are presented throughout the work. The main objective is to allow students to understand the characteristics of the OBR challenges, and to study and interact with a robot in a simulated environment. The robot behavior is implemented using a PD (Proportional Derivative) controller for line-following, and a simple fuzzy controller for obstacle avoidance.

In the last few years, there was a growth regarding the use of computational methods in the field of finance, especially to negotiations in the stock market. In this paper, we aim to bring new ideas and approaches to the development of automated trading or bots based on historical data of financial series. Our model, named Pattern Searcher, was inspired in unsupervised learning methods and evolutionary optimization. Given a trading agent with its predefined parameters, the method uses the power of Genetic Algorithm (GA) to search, within a set of financial indicators, for the region that provides a higher positive financial return. This implementation exhibited desirable properties compared to some Machine Learning methods, such as the simplification of the system flow and the generation of rules that humans can clearly understand. Besides, we have generated strategy portfolios, composed by the strategies derived from the Pattern Searcher method, which were also optimized via GA. The system was able to generate very profitable trading agents and portfolios on the Brazilian stock market, surpassing important benchmarks.

In this study we investigate the potential of using Deep Reinforcement Learning (DRL) to day trade stocks, taking into account the constraints imposed by the stock market, such as liquidity, latency, slippage and transaction costs. More specifically, we use a Deep Deterministic Policy Gradient (DDPG) algorithm to solve a series of asset allocation problems in order to define the percentage of capital that must be invested in each asset at each period, executing exclusively day trade operations. DDPG is a model-free, off-policy actor-critic method that can learn policies in high-dimensional and continuous action and state spaces, like the ones normally found in financial market environments. The proposed day trading system was tested in B3 - Brazil Stock Exchange, an important and understudied market, especially considering the application of DRL techniques to alpha generation. A series of experiments were performed from the beginning of 2017 until the end of 2019 and compared with ten benchmarks, including Ibovespa, the most important Brazilian market index, and the stock portfolios suggested by the main Brazilian banks and brokers during these years. The results were evaluated considering return and risk metrics and showed that the proposed method outperformed the benchmarks by a huge margin. The best results obtained by the algorithm had a cumulative percentage return of 311% in three years, with an annual average maximum drawdown around 19%.

Forecasting financial time series is a problem studied by researchers from different fields, who are looking for effective ways to achieve financial gains. Over time, many authors conducted studies on the possible predictability of the series through different statistical tests, and recently several papers explore the application of machine learning algorithms to have better predictions. In this paper we analyzed real data of 11 time series related to Brazilian stocks, focusing on the statistical characteristics of the series and the use of an LSTM neural network to classify future values. We analyzed the results of 5 different variance ratio tests and their relationship with the neural network classification performance. This paper proposes the application of statistical tests in the LSTM training set to highlight previously those series that have more temporal dependence and, therefore, possibly better forecast results. The results showed that 5 out of 11 stocks rejected the random walk hypothesis through the variance ratio tests and that these same stocks obtained the best performances in terms of classification and financial return.

Marine buoys aid in the battle against Illegal, Unreported and Unregulated (IUU) fishing by detecting fishing vessels in their vicinity. Marine buoys, however, may be disrupted by natural causes and buoy vandalism. In this paper, we formulate marine buoy placement as a clustering problem, and propose dropout k-means and dropout k-median to improve placement robustness to buoy disruption. We simulated the passage of ships in the Gabonese waters near West Africa using historical Automatic Identification System (AIS) data, then compared the ship detection probability of dropout k-means to classic k-means and dropout k-median to classic k-median. With 5 buoys, the buoy arrangement computed by classic k-means, dropout k-means, classic k-median and dropout k-median have ship detection probabilities of 38%, 45%, 48% and 52%., Comment: ICASSP 2020

One of the challenges of model predictive control is achieving a large domain of attraction with a small prediction horizon, in order to reduce the computation time and ease its implementation in embedded platforms. The domain of attraction can be enlarged by increasing the prediction horizon, at the expense of an increase in the number of decision variables, or by enlarging the terminal set. In MPC for tracking the terminal set is enlarged by the addition of an artificial equilibrium point as a decision variable, while maintaining stability of the closed loop system. In this paper we propose an extension of the MPC for tracking formulation that adds a single harmonic signal as an artificial reference. We show that a significant increase of the domain of attraction is achieved with the addition of a low number of decision variables, especially for low values of the prediction horizon.

The desire to live independently may conflict with safety and practical needs as old age progresses toward a state of frail elderly. With old age the risk of debilitating events rises. Having someone notice such a debilitating state and call for help becomes increasingly important. Thus, many elderly people move out of their independent residence and into a group setting, such as a nursing home, but with associated risk of depression, decline in self-confidence or self-determination. An activity monitoring system with sensors both on-body and in-home, which would detect possible debilitating events and send alerts for help, could allow some elderly people to maintain independent living. This project developed prototype modules to track room location, heart rate, and fall detection through sensors connected to a hub microcontroller system. A prototype system was developed and tested. The system shows promise but needs further development and testing before deployment.

Light field imaging technology has been attracting increasing interest because it enables capturing enriched visual information and expands the processing capabilities of traditional 2D imaging systems. Dense multiview, accurate depth maps and multiple focus planes are examples of different types of visual information enabled by light fields. This technology is also emerging in medical imaging research, like dermatology, allowing to find new features and improve classification algorithms, namely those based on machine learning approaches. This paper presents a contribution for the research community, in the form of a publicly available light field image dataset of skin lesions (named SKINL2 v1.0). This dataset contains 250 light fields, captured with a focused plenoptic camera and classified into eight clinical categories, according to the type of lesion. Each light field is comprised of 81 different views of the same lesion. The database also includes the dermatoscopic image of each lesion. A representative subset of 17 central view images of the light fields is further characterised in terms of spatial information (SI), colourfulness (CF) and compressibility. This dataset has high potential for advancing medical imaging research and development of new classification algorithms based on light fields, as well as in clinically-oriented dermatology studies.

Advances in Recommender Systems (RSs) have been focused on improving the system’s accuracy. However, accuracy alone is not enough to assess the practical effects. In real scenarios, diversity has been identified as a key dimension of recommendation utility. Thus, the main researches are focused in improve both, accuracy and diversity. This challenge remains an apparent dilemma that remains open and can boost sales by offering consumers both their mainstream and specific tastes. For this reason, we propose an approach to handle the accuracy-diversity dilemma. Our approach, based on a Particle Swarm Optimization (PSO), is a post-processing method to re-rank items from traditional RSs in order to improve diversity without accuracy losses. Experimental results in entertainment and e-commerce scenarios show that our strategy can improve users satisfaction. We improve the diversity up to 70% without significant accuracy losses.

Segmentation is a key stage in dermoscopic image processing, where the accuracy of the border line that defines skin lesions is of utmost importance for subsequent algorithms (e.g., classification) and computer-aided early diagnosis of serious medical conditions. This paper proposes a novel segmentation method based on Local Binary Patterns (LBP), where LBP and K-Means clustering are combined to achieve a detailed delineation in dermoscopic images. In comparison with usual dermatologist-like segmentation (i.e., the available ground-truth), the proposed method is capable of finding more realistic borders of skin lesions, i.e., with much more detail. The results also exhibit reduced variability amongst different performance measures and they are consistent across different images. The proposed method can be applied for cell-based-like segmentation adapted to the lesion border growing specificities. Hence, the method is suitable to follow the growth dynamics associated with the lesion border geometry in skin melanocytic images.

The world is facing a new era in which social media communication plays a fundamental role in people’s lives. Along with irrefutable benefits, several collateral drawbacks have risen, one being the wide spread of false information with malicious intents, what is now commonly called "Fake News". The fight against this problem is not easy, especially when taking into account the nature of text messages involved on social media platforms (a sea of small messages and myriad users). In this work, we cope with the challenging problem of authorship attribution of small text messages posted on social media platforms. Differently from what has been done with longer texts, we rely upon data-driven approaches, exploiting recent advances of deep neural networks in the field of pattern recognition. By viewing small texts usually employed in social media as unidimensional signals, we devise modern deep-learning techniques tailored for this kind of data to find the author of these posts with promising results.

Prof. Eric L. Adler passed away in Victoria, BC, Canada, on November 06, 2017. In this paper, Eric's biography and his contributions to the Microwave Acoustics community are highlighted, and a brief description of his influence on my personal life and career is included.

High-temperature vibration sensors are desired in a number of contemporary harsh-environment applications, namely: power plants, metallurgy, aerospace, and defense. The detection of vibrations in these applications is important for providing condition-based maintenance (CBM) for electronic, mechanical, and structural systems, thus reducing operating and maintenance costs, improving industry efficiency, and increasing safety. Microwave surface acoustic wave resonator (MSAWR) sensors have the potential of being used as high-temperature vibration sensors due to their proven harsh-environment operation, small size, battery-free, and wireless capability. This paper explores the MSA WR as a vibration sensor mounted and tested up to 500°C. A 193MHz langasite-based MSA WR sensor was attached to a custom-designed constant-stress Inconel cantilever test beam using a high-temperature capable A1 2 O 3 -based adhesive. The bam was mounted in a box furnace and exposed to vibrations using a stainless-steel rod attached to a vibration generator outside the furnace. The resulting measurements at high-temperatures reported in this paper confirm the viability of the MSA WR sensor to detect both vibration frequency and magnitude for high temperature industrial, defense, aerospace, and power sector applications.

Surface SAW resonators (SA WR) are widely used in radio-frequency (RF) communications, frequency-control, and sensing applications. External vibration-induced modulations of a SA WR's RF response can be detrimental in communications and frequency-control as it generates undesired spurious signals and unwanted phase noise. In sensing applications, vibration-induced response can either be a primary measurand or an undesired spurious signal that reduces sensor accuracy and precision. Therefore, characterization of vibration-induced modulations is important to improve device performance in all the applications listed above. This paper presents an in-phase and quadrature signal analysis technique to demodulate a SA WR RF signal exposed to external vibrations and separate the contributions of amplitude modulation (AM) and frequency modulation (FM). A SA WR was mounted in a cantilever configuration and exposed to external mechanical vibrating signals using a vibration generator. The SA WR was excited close to its resonant frequency using a RF synthesizer and its echo response was measured using a vector signal analyzer. The experimental results and findings reported in this paper confirm that the I-Q analysis developed in this work is effective not only to determine the frequency and spectral amplitude of a mechanical vibrating signal on a SA WR, but also to separate the AM and FM contributions. This analysis contributes to the quantification and understanding of induced external vibrations on SA WRs used in contemporary communications, frequency control, and sensing applications.

Navigating the trade-offs between performance, size, cost and reliability can be a challenge when considering integrated circuit (IC) packaging and the end-application. The integration of micro-electromechanical systems (MEMS), either monolithic or heterogeneous, introduces yet another level of complexity that has only recently been a major focus of multi-device packaging [1]. Wafer-level fan-out (WLFO) technology can enable improvements in several areas, primarily the reduction in size of parasitic interconnects and, proportionally, a drastic decrease in overall form-factor as compared to more ubiquitous chip-scale packaging solutions. Widespread adoption of WLFO packaging [2] has driven implementation costs to a level competitive with traditional fan-in wafer-level packaging. This study quantifies the benefits of WLFO versus flip-chip land grid array (FCLGA) packaging for a radio frequency (RF) MEMS digital tunable capacitor array integrated with 180nm CMOS technology. RF performance hinges critically upon the ability of the package to transfer signals with minimal impedance, necessitating shortened redistribution layer (RDL) paths and reduced, or removed, solder interconnects. Flip-chip packaging requires a multi-layered substrate and an intermediate solder interconnect while chip-first WLFO packaging makes use of direct Cu RDL bond to die pads and a single-level of routing to the ball grid array. Die size in both cases is 1.8x2.2x0.3mm, which enables a direct comparison of form-factor between the two package types. Manufacturability is addressed in this study; a primary challenge of monolithically integrated MEMS/CMOS is the ability to survive typical IC packaging processes wherein thermal, mechanical and electrical overstress may occur. Daisy chain packaged parts were subjected to board-level mechanical shock. Functional packaged parts were subjected to component-level reliability stresses. RF characterization of functional packaged parts was conducted on printed circuit boards (PCB); the primary figure of merit being self-resonance frequency (SRF) as a result of overall parasitic losses.

With advances on Internet of Things (IoT), network and devices management are needed due to the quantity of devices connected to transmit and collect data. IoT management presents the challenge to control and monitoring heterogeneous network elements, physical or logical, that support different protocols and quality of service (QoS) requirements. Devices interoperability and heterogeneity are key management challenges in IoT scenarios. IoT network management has an issue regarding the diversity of protocols and platforms where a comparative evaluation is difficult to perform due to this lack of standardization. This paper presents a performance evaluation study of open source IoT management platforms using the most popular IoT network management protocols considering real environments. For comparison purposes, the real environment of a smart lighting system is used. The evaluation study considers both qualitative and quantitative metrics. Through the collected data, OpenDayLight (ODL) platform presents the best results considering the use of NETCONF and simple network management protocol (SNMP). It natively supports both IoT network and IoT devices management protocols. It is concluded that ODL platform is a modular platform where it is possible configuring the network management protocols appropriately for each scenario, either it is a scenario that requires real-time management and energy saving or a scenario with low response time and large transfer rate.

Nowadays, it is possible to note many machine learning techniques being applied to predict financial time series. However, recent studies indicate that the performance of such techniques can be strongly affected by data representation. In this manuscript we propose a combination of two machine learning algorithms to detect trends in stock market prices. In this approach, Boltzmann Restricted Machines are used as the latent feature extractor and Support Vector Machines work as the classifier. We performed tests with real data of five assets from the Brazilian Stock Market, BM&FBOVESPA. The results obtained with the proposed combination were better when compared to those ones reached by Support Vector Machines only. This suggests that the proposed approach can be suitable for the considered application.

Before the advent of computers and Internet, the stock market investors perform their operations based mainly on intuition. With the growth of investments and online stock trading, a continued search for better tools to improve the prediction of stock market trends has become necessary in order to increase profits and reduce risks. In this work we propose and evaluate some algorithmic trading (algotrading) strategies, based on an Ensemble of artificial neural networks (ANN), to support the decision of stock market's investors. Thirty different ANN models, using different input sets of price, volume and technical indicators, were analyzed for different stock symbols (i.e., companies from different economy sectors) of the main Brazilian Stock Market-BM& FBovespa. Moreover, ensembles that combine the best ANN models were modeled and validated through different experiments. The results confirm the potential of the proposed strategies for algotrading.

Internet of Things (IoT) is a paradigm where devices are connected to the Internet and can interact with them, performing a machine-to-machine (M2M) communication. One of the challenges of IoT is related to the heterogeneity of devices given their different characteristics, like processing capacity and communication technologies. It demands a dynamic and a context-aware configuration management system that meets the requirements of the networks and their various functionalities. This paper focuses on the performance evaluation of the simple network management protocol (SNMP) and NETCONF in a real environment using an IoT-based smart lighting system. Among the available IoT network management open source platforms, OpenDayLight was chosen because it supports both SNMP and NETCONF. The following performance metrics are used: average response time, user request errors, throughput, and data rate. It was observed that NETCONF performs better than SNMP in the studied scenario. It was concluded that NETCONF becomes more viable in a scenario that requires high IoT network management traffic (non-real time) because it takes less time to get response messages with higher throughput. But in a scenario that needs real-time management and energy savings, SNMP performs better because its frames contain fewer bytes than NETCONF turning it a more straightforward protocol.

Which distribution commonly used to fit a species abundance distribution gives the best fit across several area sizes? We approach this problem by looking at the moments of the distribution and compare these with those the distribution based on empirical data on tree species collected in Barro Colorado Island, Panama. We conclude that no single distribution gives the best fit at all scales.

During the image acquisition process, some level of noise is usually added to the real data mainly due to physical limitations of the acquisition sensor, and also regarding imprecisions during the data transmission and manipulation. Therefore, the resultant image needs to be processed in order to attenuate its noise without loosing details. Machine learning approaches have been successfully used for image denoising. Among such approaches, Restricted Boltzmann Machine (RBM) is one of the most used technique for this purpose. Here, we propose to enhance the RBM performance on image denoising by adding a posterior supervision before its final denoising step. To this purpose, we propose a simple but effective approach that performs a fine-tuning in the RBM model. Experiments on public datasets corrupted by different levels of Gaussian noise support the effectiveness of the proposed approach with respect to some state-of-the-art image denoising approaches.

The uptake of electric vehicles (EV) will require important modifications in traditional grid planning and load forecasting techniques. Existing literature suggests that the integration of EVs will be more adversarial to elements of the existing electricity infrastructure in terms of power supply (kW) than energy (kWh) delivery. While several studies analyzed the grid impact of electric vehicle fleets, few consider the adoption process itself which may lead to strong spatial variations of the utilization of charging infrastructure. The presented approach extends spatial load forecasting, introducing diffusion theory elements to analyze spatio-temporal clustering of EV charging demand. Using open-access census and grid data, this work develops a deterministic framework to forecast spatial patterns of EV charging applied to a real-world environment. Outcomes suggest substantial spatial clustering of EV adoption patterns, showing substation overrating for EV penetration rates of 25% and above with 7.4kW charging power.

Surface acoustic wave (SAW) devices have been shown recently to be an appropriate technology for harsh-environment sensor applications, such as those in the automotive, aerospace, power plant, and metallurgic industries. Usages include temperature, pressure, torque, strain, gas, liquid, biochemical, and mass sensors in structural health monitoring, industrial process monitoring, environment checking and control, and condition based maintenance. For harsh-environment high-temperature applications, battery-free wireless operation associated with sensor robustness, small profile, and mass fabrication capability make SAW sensors a very attractive solution. For high-temperature ( 2 O 3 -based electrode, with atomic layer deposited (ALD) Al 2 O 3 protective layer and yttria-stabilized zirconia (YSZ) gas sensing layer. The effect of 50nm thick Al 2 O 3 or YSZ layers on SAWR sensor resonant frequency and temperature sensitivity responses were investigated. Temperature sensitivities of SAWR sensors with and without ALD Al 2 O 3 protective layer have been compared with numerical predictions for a bare substrate. Measured temperature sensitivities of SAWR sensors with and without the YSZ sensing layers have been compared against each other. In addition, the stability of the SAWR sensor frequency responses have been monitored during multiple 110-hour temperature cycling (over 620 hours in total) from 300°C to 750°C. The measured frequency deviation at 750°C over six of these 110-hour cycling tests was 0.05%. The results obtained confirm that the presence of the ALD Al 2 O 3 protective layer or the YSZ sensing layer do not compromise the stability of high-temperature harsh-environment SAWR sensors, validating their use for the aforementioned applications.

In this paper, a two-step methodology is presented to design a low power wireless energy harvesting device. The rectenna system, consisting of an half-wave rectifier circuit and a non-50 ohms antenna, is designed to archive maximal power transfer at some specified frequency and power level. The main idea of this work is to obtain maximum efficiency avoiding complex impedance matching circuits. First, the rectifier circuit is designed and its input impedance is measured under the desired operating conditions (frequency range and input power levels). Then, the antenna geometry is optimized to match the complex impedance of the rectifier. Results are presented for a 2.45 GHz rectenna. The performance of the non-50 ohms antenna is compared to a conventional one.

We present our recent work on the Weyl-Heisenberg ensemble and its statistical properties [4]. The WH ensemble is a class of determinantal point processes associated with the Schrodinger representation of the Heisenberg group. As a special example, WH ensembles include a multi-layer extension of the Ginibre ensemble modeling the distribution of electrons in higher Landau levels. We describe the hyperuniformity of WH ensembles, which characterizes a state of matter for which (scaled) density fluctuations diminish towards zero at the largest length scales. Our approach is based on methods from time-frequency analysis. We introduce the main results from [4] highlighting time-frequency techniques and connections to the theory of polyanalytic functions, and also present some small extensions.

Large volumes of data are generated and collected nowadays through sensors, devices, social networks, among others. The ability to handle large datasets has become an important expertise for the success of many organizations around the world, demanding increasingly parallel and distributed processing. Parallel computing, based on multicore architecture, re-emerged in recent years as a solution for this challenge. To help developers to design parallel programs, there exists several parallel and distributed tools (frameworks), such as Apache Hadoop and Spark. These frameworks provide a plenty of configuration parameters (e.g., Hadoop has more than 200) and to configure all of them optimally is a non-trivial task. This work investigates the influence of various parameters on the performance of Apache Hadoop when process HEDA, an iterative algorithm that calculate metrics of centrality in large graphs. The execution of HEDA in a complex network is extremely important because there are several centrality measures that determine the importance of a node in the graph. The centrality of a node is a key issue in the analysis of complex networks, especially on social networks. It was observed that in some cases the improvement in execution time reached almost 80% by applying the proposed values in the selected parameters. Moreover, it was possible to increase CPU consumption four times and to achieve a Speedup of 92% of linearity. The work also describes the steps that were followed to perform the experiments, which can help other researchers to conduct new experiments.

Data from the survey on patients' perspectives on electronic health records requirements and barriers for adoption in physiotherapy are presented and discussed in view of tailoring information and communication technologies (ICTs) to support physiotherapy for rural elderly. Socio-demographic data on rural elderly suggest that they could not be aware or understand the ICTs based healthcare tools that may improve accessibility and quality of care. By integrating personally controlled health records, patient reported outcomes and educational resources, the understanding and efficient use of health information technologies may increase and medical errors may be prevented.

The detection of dissolved carbon dioxide (dCO 2 ) is made possible through a colorimetric effect that occurs in a sensitive membrane. The reaction with dCO 2 changes the pH of the membrane causing a small difference in its colour which results in a characteristic absorbance spectrum band near 435 nm. A sensing platform based on this effect was developed and tested in gaseous and in aqueous environments. It is a combination of a bundle of large core fibre optics (with diameters above 200 μm) with light emission diodes (LEDs) in the visible range of the spectrum, a silicon photodetector and a polymer membrane sensitive to CO 2 . A variation in the absorption of 3 / %VV was obtained in the range from 0 to 1.6 % of gaseous CO2 with an estimated response time below 60 seconds.

After showing in a previous work that PD measurements suffer from poor repeatability and reproducibility, the present study turns now to a more qualitative approach, investigating if a method of PD fault location along a Roebel bar length can be used to identify insulation critical areas that later on could fail under electrical stress. Typical insulation “damages” have been created on the components in advance by submitting them to both thermal-mechanical and electrical aging tests. Once such preliminary aging was done, the critical areas along the bars length were identified by mapping the PD readings on each component. Afterwards, every bar was subjected to a destructive electrical breakdown test. At last, the breakdown locations were compared with the PD mappings to search for any possible correlation between the two events. Based upon the present results, such correlation does not exist.

Predictions on stock market prices are a great challenge due to the fact that it is an immensely complex, chaotic and dynamic environment. There are many studies from various areas aiming to take on that challenge and Machine Learning approaches have been the focus of many of them. There are many examples of Machine Learning algorithms been able to reach satisfactory results when doing that type of prediction. This article studies the usage of LSTM networks on that scenario, to predict future trends of stock prices based on the price history, alongside with technical analysis indicators. For that goal, a prediction model was built, and a series of experiments were executed and theirs results analyzed against a number of metrics to assess if this type of algorithm presents and improvements when compared to other Machine Learning methods and investment strategies. The results that were obtained are promising, getting up to an average of 55.9% of accuracy when predicting if the price of a particular stock is going to go up or not in the near future.

The Walsh-Hadamard transform plays a major role in many image and video coding algorithms. In one hand, its intensive use in these algorithms makes its acceleration a challenge, in order to speed-up the algorithm execution. On the other hand, the available fast implementations are not efficient across different platforms. In this work, a parallel-based implementation of the WHT is proposed for CPU and GPU platforms using the OpenCL standard. OpenCL achieves portability at code level, but its performance suffers when the same code is used for CPUs and GPUs. To achieve top performance, we propose two WHT versions: OpenCL-GPU for GPUs and OpenCL-CPU for CPUs. Broadly, OpenCL-GPU executed on a GPU runs faster than OpenCL-CPU executed on a multicore CPU, with speedups that range from 120.87 to 1016.35. However, OpenCL-GPU performance drops substantially when ran on a multicore CPU machine, where OpenCL-CPU achieves higher performance, as it exploits the OpenCL support for SIMD instructions.

Strain sensors are used to guarantee integrity, safety, and appropriate maintenance in a multitude of applications such as civil structures, industrial equipment, and aircrafts. In harsh environments associated with power plants, aerospace, oil drilling, and industrial processes, there is a pressing need for robust, reliable, small footprint strain sensors, which can operate wirelessly, battery-free, and require minimal or no maintenance. The University of Maine group has been reporting on successful harsh-environment surface acoustic wave resonator (SAWR) temperature sensors fabricated on langasite (LGS) substrates and employing high temperature Pt-based electrode alloys. The usage of this platform for strain sensor application poses technological challenges, such as packaging and sensor assembly for the temperature span; packaged sensor stability and longevity over multiple cycles; and finally temperature cross-sensitivity. This paper discusses the design, fabrication, calibration, and testing up to 400°C of LGS SAWR static strain sensors using co-deposited Pt-Al 2 O 3 and PtNi-PtZr multi-layered electrodes operating at 195 MHz and 338 MHz. The LGS SAWR sensors were mounted on a constant stress beam and characterized up to 200 microstrains, μe. At room temperature, a commercial strain gauge was used for calibration. At higher temperatures (up to 400°C), the sensors were calibrated against a finite element analysis (FEA) model, due to the inexistency of reliable static commercial sensors at those temperatures. Linear regression of the SAWR frequency shift versus strain resulted in correlations of 0.997 and 0.976 at 300°C and 400°C, respectively. The associated extracted sensor sensitivities to strain at 300°C and 400°C were 42Hz/μe and 37Hz/μe. The reported results confirmed the viability of using LGS SAWR as high temperature static strain sensors.