Power efficient refined seizure prediction algorithm based on an enhanced benchmarking

Deep learning techniques have led to significant advancements in seizure prediction research. However, corresponding used benchmarks are not uniform in published results. Moreover, inappropriate training and evaluation processes used in various work create overfitted models, making prediction performance fluctuate or unreliable. In this study, we analyzed the various data preparation methods, dataset partition methods in related works, and explained the corresponding impacts to the prediction algorithms. Then we applied a robust processing procedure that considers the appropriate sampling parameters and the leave-one-out cross-validation method to avoid possible overfitting and provide prerequisites for ease benchmarking. Moreover, a deep learning architecture takes advantage of a one-dimension convolutional neural network and a bi-directional long short-term memory network is proposed for seizure prediction. The architecture achieves 77.6% accuracy, 82.7% sensitivity, and 72.4% specificity, and it outperforms the indicators of other prior-art works. The proposed model is also hardware friendly; it has 6.274 k parameters and requires only 12.825 M floating-point operations, which is advantageous for memory and power constrained device implementations.

Analysis of synchronous magnetotelluric and magnetovariational regime observations for the Kentor mini test polygon

The paper discusses various approaches for processing and analysis of synchronous magnetotelluric and magnetovariational data obtained at the Kentor mini test polygon, which is located in the Baytic basin of the Chui region of Kyrgyzstan. The versions of noise reduction of received electromagnetic signals at the stage of robust processing are considered. The materials were processed using a remote reference technique for possible noise detection and calculate the additional components, such as a horizontal magnetic tensor for further interpretation of the obtained data. The dynamics of changes in the geoelectric cross-section for 2 sessions of researches by the tested profiles is considered.

Prediction of Mechanical Properties of the Stirrup-Confined Rectangular CFST Stub Columns Using FEM and Machine Learning

In this study, a machine learning method using gradient boost regression tree (GBRT) model was presented to predict the ultimate bearing capacity of stirrup-confined rectangular CFST stub columns (SCFST) by using a comprehensive data set and by adjusting the selected parameters indicated in the previous research (B, D, t, ρsa, fcu, fs). The advantage of GBRT is its strong predictive ability, which can naturally handle different types of data and very robust processing of outliers out of space. The comprehensive data set obtained from the FEM method which has been verified the accuracy and rationality by the existing literature. In order to make the data group closer to the engineering example, a large amount of experimental data collected in the literature was added to the data group to enhance the accuracy of the model. We compare a few regression models simply and the results show that the GBRT model has a good predictive effect on the mechanical properties of CFST columns. In summary, it can help pre-investigations for the CFST columns.

Studi Magnetotelurik (MT) untuk Mendelineasi Potensi Regional Gas Serpih Bawah Permukaan Berdasarkan Properti Tahanan Jenis di Cekungan Kutai, Kalimantan Timur

Telah dilakukan penelitian menggunakan metode magnetotelurik (MT) pada 40 titik lokasi pengukuran di sekitar Cekungan Kutai, Kalimantan Timur. Studi ini bermaksud untuk mengkarakterisasi keberadaan batuan induk untuk potensi gas serpih berdasarkan nilai tahanan jenis. Guna memperoleh gambaran kondisi geologi bawah permukaan yang baik, pengolahan data secara objektif dilakukan agar memperoleh kualitas data MT terbaik. Beberapa teknik pengolahan data seperti robust processing dan seleksi cross power (XPR) diaplikasikan untuk memisahkan komponen data dari gangguan sehingga memperoleh gambaran bawah permukaan yang dapat dipercaya dari kedalaman yang relatif dangkal hingga sangat dalam melalui tahap pemodelan inversi 2-D. Model inversi terdiri dari 3 penampang vertikal, yaitu penampang A-B, C-D dan E-F dengan arah relatif barat – timur yang memotong Antiklinorium Samarinda dengan panjang lintasan berturut-turut ±50 km, ±48 km dan ±32 km. Berdasarkan ketiga penampang vertikal tersebut, informasi sebaran anomali tahanan jenis rendah regional berhasil dipetakan yang selanjutnya ditafsirkan sebagai respon keberadaan batuan induk gas serpih di bawah permukaan. Pada penampang A-B, bagian atas lapisan konduktif ini diperoleh pada kedalaman yang bervariasi pada rentang kedalaman 1000-5000 m di bawah permukaan. Sementara itu, lapisan konduktif pada penampang C-D diperoleh pada kedalaman yang bervariasi dengan rentang 1500-7000 m di bawah permukaan, sedangkan anomali serupa pada penampang E-F diperoleh pada sekitar kedalaman 2000-3000 m. Variasi kedalaman lapisan konduktif ini ditafsirkan akibat adanya struktur geologi berupa patahan dan lipatan di sepanjang ketiga penampang vertikal yang tercermin dari adanya Antiklinorium Samarinda di permukaan.Katakunci: Cekungan Kutai, gas serpih, magnetotelurik, tahanan jenis.

Parametric Control of Melting Gel Morphology and Chemistry via Electrospray Deposition

Melting gels are a class of hybrid organic-inorganic, silica-based sol-gels which are solid below their glass transition temperatures, near room temperature, but show thermoplastic behavior when heated. While this phase change can be repeated multiple times, heating the gel past its consolidation temperature, typically above 130 °C, initiates an irreversible reaction that produces highly crosslinked glassy organic/inorganic materials via hydrolysis and polycondensation. This ability makes melting gels uniquely compatible with processing techniques inaccessible to other sol-gels. By properly tuning their properties, it should be possible to create protective coatings for electronics and anti-corrosive coatings for metals that are highly hydrophobic and insulating. However, melting gel consolidation reactions are highly dependent on charge interactions, raising the question of how these materials will respond to a processing technique, like electrospray deposition (ESD), which is dependent on charge delivery. In this study, we focus on the role that substrate temperature and charge polarity play on film morphology, consolidation chemistry, and surface properties when processing via ESD. Optical images, film thickness measurements, and FTIR were used to characterize the sprayed melting gel with the goal of developing a robust processing space for producing highly cross linked, hydrophobic, dielectric coatings.

Development of an Optical Method for the Evaluation of Whole Blood Coagulation

Blood coagulation is a defense mechanism, which is activated in case of blood loss, due to vessel damage, or other injury. Pathological cases arise from malfunctions of the blood coagulation mechanism, and rapid growth of clots results in partially or even fully blocked blood vessel. The aim of this work is to characterize blood coagulation, by analyzing the time-dependent structural properties of whole blood, using an inexpensive design and robust processing approaches. The methods used in this work include brightfield microscopy and image processing techniques, applied on finger-prick blood samples. The blood samples were produced and directly utilized in custom-made glass microchannels. Color images were captured via a microscopy-camera setup for a period of 35 min, utilizing three different magnifications. Statistical information was extracted directly from the color components and the binary conversions of the images. The main advantage in the current work lies on a Boolean classification approach utilized on the binary data, which enabled to identify the interchange between specific structural elements of blood, namely the red blood cells, the plasma and the clotted regions, as a result of the clotting process. Coagulation indices produced included a bulk coagulation index, a plasma-reduction based index and a clot formation index. The results produced with the inexpensive design and the low computational complexity in the current approach, show good agreement with the literature, and a great potential for a robust characterization of blood coagulation.

A PRACTICAL APPROACH FOR SELF-POTENTIAL DATA ACQUISITION, PROCESSING, AND VISUALIZATION

We propose a comprehensive methodology for the acquisition and processing of self-potential (SP) data, as well as some keys to the interpretation of the results. The wide applicability of the SP method, and its low cost, make it a popular method for use in a variety of natural environments. Despite its versatility and the fact that various published journal papers describe the method and its applications, we believe that there is an important need for a dedicated, peer-reviewed SP acquisition, processing and visualization/interpretation paper in the scientific literature. We identified a great interest from the scientific community for such a journal paper as a guide for both existing and new practitioners with their SP survey design, data acquisition, robust processing, and initial interpretation steps. A step-by-step methodology is proposed here for SP data acquisition and processing, combined with practical guidance for the interpretation of collected and processed SP data, including a discussion of common errors and typical sources of uncertainty. The presented examples are based on studies in volcanic environments (e.g. hydrothermal systems), however the processing steps and methodology are fully applicable and transferable across disciplines to SP data acquired in any environment, and for a wide variety of applications. After a short overview of the field acquisition method and the low-cost equipment, the reference and closure corrections, their meaning for the SP signal, and their effect on the dataset are detailed and exemplified. The benefits of interpolating SP data in two steps is discussed. Combining map visualization, SP vs distance, and SP vs elevation graphs appears as a highly effective strategy to interpret the signal in terms of hydrogeological and hydrothermal domains, and to highlight structural limits in volcanic contexts as well as in other environments.