Accelerating Machine-Learning Algorithms on FPGAs using Pattern-Based Decomposition

Machine-learning algorithms are employed in a wide variety of applications to extract useful information from data sets, and many are known to suffer from super-linear increases in computational time with increasing data size and number of signals being processed (data dimension). Certain principal machine-learning algorithms are commonly found embedded in larger detection, estimation, or classification operations. Three such principal algorithms are the Parzen window-based, non-parametric estimation of Probability Density Functions (PDFs), K-means clustering and correlation. Because they form an integral part of numerous machine-learning applications, fast and efficient execution of these algorithms is extremely desirable. FPGA-based reconfigurable computing (RC) has been successfully used to accelerate computationally intensive problems in a wide variety of scientific domains to achieve speedup over traditional software implementations. However, this potential benefit is quite often not fully realized because creating efficient FPGA designs is generally carried out in a laborious, case-specific manner requiring a great amount of redundant time and effort. In this paper, an approach using pattern-based decomposition for algorithm acceleration on FPGAs is proposed that offers significant increases in productivity via design reusability. Using this approach, we design, analyze, and implement a multi-dimensional PDF estimation algorithm using Gaussian kernels on FPGAs. First, the algorithm’s amenability to a hardware paradigm and expected speedups are predicted. After implementation, actual speedup and performance metrics are compared to the predictions, showing speedup on the order of 20× over a 3.2 GHz processor. Multi-core architectures are developed to further improve performance by scaling the design. Portability of the hardware design across multiple FPGA platforms is also analyzed. After implementing the PDF algorithm, the value of pattern-based decomposition to support reuse is demonstrated by rapid development of the K-means and correlation algorithms.     

Thermal stability and transport studies of (100 − 2x)TeO2-xAg2O-xWO3 (7.5 ≤ x ≤ 30) glass system

Differential scanning calorimetry (DSC), infrared (IR) and direct current (DC) conductivity studies have been carried out on (100 − 2x)TeO₂-xAg₂O-xWO₃ (7.5 ≤ x ≤ 30) glass system. The IR studies show that the structure of glass network consists of [TeO₄], [TeO₃]/[TeO₃₊₁], [WO₄] units. Thermal properties such as the glass transition (T_g), onset crystallization (T_o), thermal stability (ΔT), glass transition width (ΔT_g), heat capacities in the glassy and liquid state (C_pg and C_pl), heat capacity change (ΔC_p) and ratios C_pl/C_pg of the glass systems were calculated. The highest thermal stability (237 °C) obtained in 55TeO₂-22.5Ag₂O-22.5WO₃ glass suggests that this new glass may be a potentially useful candidate material host for rare earth doped optical fibers. The DC conductivity of glasses was measured in temperature region 27-260 °C, the activation energy (E_act) values varied from 1.393 to 0.272 eV and for the temperature interval 170-260 °C, the values of conductivity (σ) of glasses varied from 8.79 × 10⁻⁹ to 1.47 × 10⁻⁶ S cm⁻¹.     

Composite polymeric microsponge-based long-acting gel formulation for topical delivery of mupirocin

Topical delivery of medicaments in a controlled manner is still a promising area of research. Drug-containing dammar gum-ethyl cellulose composite microsponge loaded gel formulation (D-MSPG) was developed for controlled topical delivery of mupirocin. The drug-loaded microsponges (D-MSPs) were formulated by the quasi-emulsion solvent diffusion method and were evaluated for morphology, particle size distribution, entrapment efficiency, thermal properties, and crystallinity. The optimized D-MSPs (entrapment efficiency 91.5 ± 4.0% and particle size of 55.15 ± 2.9 μm) were dispersed in carbopol 934 gel (D-MSPG). The final product was characterized for pH, viscosity, texture, spreadability, consistency, syneresis, in vitro drug release, and ex vivo skin penetration study. A comparative study with marketed formulation was performed. For optimized gel formulation (G4), drug content was 104.19 ± 1.68%, and drug release was 84.19% after 24 h. The pH of the optimized gel was observed to be 6.05 ± 0.04. Viscosity of the optimized gel formulation was found to be 1212.15 ± 434.85 mPa-s at 50 s⁻¹. The steady-state flux (J) in ex vivo skin permeation was observed to be 53.96 μg cm⁻² h⁻¹ and the permeability coefficient was 2.69 cm/h for the optimized gel formulation. According to the findings, the D-MSPG-based formulation strategy can act well to prolong the topical delivery of mupirocin or similar drug molecules.     

An integrated design for improving spectral efficiency of FSO communication system in 6G applications

In the case of future access networks, such as 6G-based applications, the free-space optical communication (FSO) technology is an efficient solution. FSO in 6G is extremely popular because of its unique properties, which include ease of installation, high bandwidth, high security, license-free long range, and interference resistance. However, environmental disturbances have a negative impact on the FSO system's functioning. Because of these atmospheric turbulences, the optical information gets restricted, which in turn reduces the link reliability, power density, and distance. This paper proposes a hybrid FSO strategy for improving system efficiency to address these problems. Signal creation is fundamental to every successful communication system. The FSO system consists of three components: the transmitter, the channel medium, and the receiver. The transmitter section includes a continuous wave laser, a Mach–Zehnder modulator, a nonreturn to zero transpose Walsh code generator, and a nonreturn to zero pulse generator. Wavelength division multiplexing-optical code division multiple access-spectrum slicing (WDM-OCDMA-SS) is used to facilitate efficient data transfer after signals have been generated by alternate mark inversion (AMI). In order to boost a weak signal, fused Raman erbium amplifiers (Fuse-RE) are used. Q factor and log bit error rate (BER) are used to assess the performance of the suggested approach. The proposed model has obtained a log BER of −15.3291 for clear air and a Q factor of 8.2922, whereas the performances are implemented using Python. The proposed approach achieves better performance when compared to the existing methodologies.     

Ligand-Specific Nano-Contrast Agents Promote Enhanced Breast Cancer CT Detection at 0.5 mg Au

For many cancer types, being undetectable from early symptoms or blood tests, or often detected at late stages, medical imaging emerges as the most efficient tool for cancer screening. MRI, ultrasound, X-rays (mammography), and X-ray CT (CT) are currently used in hospitals with variable costs. Diagnostic materials that can detect breast tumors through molecular recognition and amplify the signal at the targeting site in combination with state-of-the-art CT techniques, such as dual-energy CT, could lead to a more precise detection and assist significantly in image-guided intervention. Herein, we have developed a ligand-specific X-ray contrast agent that recognizes α5β1 integrins overexpressed in MDA-MB-231 breast cancer cells for detection of triple (−) cancer, which proliferates very aggressively. In vitro studies show binding and internalization of our nanoprobes within those cells, towards uncoated nanoparticles (NPs) and saline. In vivo studies show high retention of ~3 nm ligand-PEG-S-AuNPs in breast tumors in mice (up to 21 days) and pronounced CT detection, with statistical significance from saline and iohexol, though only 0.5 mg of metal were utilized. In addition, accumulation of ligand-specific NPs is shown in tumors with minimal presence in other organs, relative to controls. The prolonged, low-metal, NP-enhanced spectral-CT detection of triple (−) breast cancer could lead to breakthrough advances in X-ray cancer diagnostics, nanotechnology, and medicine.     

Solution‐Processed Small‐Molecule Bulk Heterojunction Ambipolar Transistors

Solution‐processed small‐molecule bulk heterojunction (BHJ) ambipolar organic thin‐film transistors are fabricated based on a combination of [2‐phenylbenzo[d,d']thieno[3,2‐b;4,5‐b']dithiophene (P‐BTDT) : 2‐(4‐n‐octylphenyl)benzo[d,d ']thieno[3,2‐b;4,5‐b']dithiophene (OP‐BTDT)] and C₆₀. Treating high electrical performance vacuum‐deposited P‐BTDT organic semiconductors with a newly developed solution‐processed organic semiconductor material, OP‐BTDT, in an optimized ratio yields a solution‐processed p‐channel organic semiconductor blend with carrier mobility as high as 0.65 cm² V^?1 s^?1. An optimized blending of P‐BTDT:OP‐BTDT with the n‐channel semiconductor, C₆₀, results in a BHJ ambipolar transistor with balanced carrier mobilities for holes and electrons of 0.03 and 0.02 cm² V^?1 s^?1, respectively. Furthermore, a complementary‐like inverter composed of two ambipolar thin‐film transistors is demonstrated, which achieves a gain of 115.     

Redefining the Diamond Cutting Edge: A Technique that Complements Nano-Metric Surface Generation

Ongoing developments in broadband digital networks and optical devices require mechanical components having nano-metric surface finishes and ultra precision shapes. Such components are processed using the technology of diamond micro-machining in which diamond tools are extensively used. However, these tools are susceptible to chemical attack at high temperatures and induce severe wear, especially when cutting ferrous materials. As a result, these diamond tools need to be redefined (i.e., resharpened) on a regular basis in order to facilitate micro-cutting and to generate nano-metric surface finishes. This paper describes a new way of diamond tool edge re-sharpening and its conditions to achieve both increased accuracy and material removal rate.     

Novel thermo‐responsive semi‐interpenetrating network microspheres of gellan gum‐poly(N‐isopropylacrylamide) for controlled release of atenolol

Thermoresponsive microspheres of gellan gum‐poly(N‐isopropylacrylamide), i.e., GG‐P(NIPAAm) semi‐interpenetrating polymer networks (semi‐IPNs) have been prepared by ionic crosslinking and used to study the controlled release (CR) of atenolol (ATL), an antihypertensive drug. Interaction of the drug with polymers was studied by Fourier transform infrared (FTIR) spectroscopy. Differential scanning calorimetry (DSC) was used to confirm the polymorphism and molecular level dispersion of ATL. Scanning electron microscopy (SEM) indicated spherical nature and smooth surfaces of the microspheres with some debris attached on their surfaces. Mean particle size measured by laser light diffraction ranged between 34 and 76 μm. Equilibrium swelling performed at 25°C and 37°C in pH 7.4 phosphate buffer exhibited thermoresponsive nature of the polymers. In vitro drug release performed at 25°C and 37°C indicated temperature‐dependency of ATL release, which was extended up to 12 h. In vitro release profiles at both the temperatures confirmed thermoresponsive nature of the polymers giving pulsatile trends. The % cumulative release data have been fitted to an empirical equation to estimate transport parameters and to understand the nature of drug release. © 2010 Wiley Periodicals, Inc. J Appl Polym Sci, 2010     

Chemically functionalized carbon nanotubes and their characterization using thermogravimetric analysis, fourier transform infrared, and raman spectroscopy

This article reports key findings on the chemical functionalization of carbon nanotubes (CNT). The functionalization of chemical vapor-deposited CNT was carried out by treating tubes with polyvinyl alcohol through ultrasonication in water with the aid of a surfactant. The surfactant is expected to promote the unbundling of aggregated CNT. The characterization of functionalized samples using thermogravimetric analysis, Fourier transform infrared spectroscopy, and Raman spectroscopy revealed that the CNT were functionalized by the interaction of carboxylic acid and hydroxyl groups. From the characterization studies, it is apparent that there is a strong interaction between these functional groups and the covalently bonded carbon in the CNT network. The functionalization process enabled good CNT dispersion in the solution, and the CNT remained in suspension for many days. To support the effective functionalization of the tubes, the interaction of functionalized CNT with Ni ions is also demonstrated. This paper was presented at the fourth International Surface Engineering Congress and Exposition held August 1–3, 2005 in St. Paul, MN.     

TEM and high-temperature X-ray diffractometric studies on the structural transformations in Ni3Al

High-temperature X-ray diffraction and TEM studies were conducted on nickel-rich boron-doped Ni₃Al in order to confirm our earlier observations on the existence of a structural transformation in these alloys. The results obtained are discussed through a model proposed. The L1₂ structure appears to transform to another L1₂ or to a DO₂₂ structure during heating. Such a transformation starts at around 700 °C and seems to complete around 1100 °C and appears to be of a continuous type. In the temperature range 700–1100 °C both phases coexist, which causes a tetragonal distortion of the L1₂ lattice giving rise to a tweed morphology in TEM observations. The transformation mechanism involves a periodic modulation of lattice constant. The calculated values for this periodicity (expressed as a number of unit cells in the [100] direction) obtained from X-ray diffraction (L = 62, 70, 74) matched well with that (L = 65) obtained using TEM.