Fading memory as inductive bias in residual recurrent networks.

Residual connections have been proposed as an architecture-based inductive bias to mitigate the problem of exploding and vanishing gradients and increased task performance in both feed-forward and recurrent networks (RNNs) when trained with the backpropagation algorithm. Yet, little is known about how residual connections in RNNs influence their dynamics and fading memory properties. Here, we introduce weakly coupled residual recurrent networks (WCRNNs) in which residual connections result in well-defined Lyapunov exponents and allow for studying properties of fading memory.

A novel feature-scrambling approach reveals the capacity of convolutional neural networks to learn spatial relations.

Convolutional neural networks (CNNs) are one of the most successful computer vision systems to solve object recognition. Furthermore, CNNs have major applications in understanding the nature of visual representations in the human brain. Yet it remains poorly understood how CNNs actually make their decisions, what the nature of their internal representations is, and how their recognition strategies differ from humans.

Modern Artificial Neural Networks: Is Evolution Cleverer?

Machine learning tools, particularly artificial neural networks (ANN), have become ubiquitous in many scientific disciplines, and machine learning-based techniques flourish not only because of the expanding computational power and the increasing availability of labeled data sets but also because of the increasingly powerful training algorithms and refined topologies of ANN. Some refined topologies were initially motivated by neuronal network architectures found in the brain, such as convolutional ANN. Later topologies of neuronal networks departed from the biological substrate and began to be developed independently as the biological processing units are not well understood or are not transferable to in silico architectures.

Crystalline and magnetic properties of CoO nanoparticles locally investigated by using radioactive indium tracer.

We herein report a comprehensive investigation on the magnetic, structural, and electric properties of CoO nanoparticles with different sizes by local inspection through hyperfine interactions measured in a wide range of temperatures (10-670 K) by using radioactive [Formula: see text]In([Formula: see text]Cd) tracers with the perturbed angular correlations technique. Small cobalt oxide nanoparticles with the characteristic size of 6.5 nm have been prepared by the wet chemical route that turned out to be essential to incorporate radioactivity tracers during nucleation and growth of the particles.

Ray-tracing simulation of the radiation dose distribution on the surface of the spherical phantom of the MATROSHKA-R experiment onboard the ISS.

Space radiation is one of the main concerns for human space flights. The prediction of the radiation dose for the actual spacecraft geometry is very important for the planning of long-duration missions. We present a numerical method for the fast calculation of the radiation dose rate during a space flight.

A regularity index for dendrites - local statistics of a neuron's input space.

Neurons collect their inputs from other neurons by sending out arborized dendritic structures. However, the relationship between the shape of dendrites and the precise organization of synaptic inputs in the neural tissue remains unclear. Inputs could be distributed in tight clusters, entirely randomly or else in a regular grid-like manner.

Universal features of dendrites through centripetal branch ordering.

Dendrites form predominantly binary trees that are exquisitely embedded in the networks of the brain. While neuronal computation is known to depend on the morphology of dendrites, their underlying topological blueprint remains unknown. Here, we used a centripetal branch ordering scheme originally developed to describe river networks-the Horton-Strahler order (SO)-to examine hierarchical relationships of branching statistics in reconstructed and model dendritic trees.

Economically attractive route for the preparation of high quality magnetic nanoparticles by the thermal decomposition of iron(III) acetylacetonate.

The thermal decomposition (TD) methods are among the most successful in obtaining magnetic nanoparticles with a high degree of control of size and narrow particle size distribution. Here we investigated the TD of iron(III) acetylacetonate in the presence of oleic acid, oleylamine, and a series of alcohols in order to disclose their role and also investigate economically attractive alternatives for the synthesis of iron oxide nanoparticles without compromising their size and shape control. We have found that some affordable and reasonably less priced alcohols, such as 1,2-octanediol and cyclohexanol, may replace the commonly used and expensive 1,2-hexadecanediol, providing an economically attractive route for the synthesis of high quality magnetic nanoparticles.

Binary prefix for sampling frequency offset estimation in dispersive optical transmissions.

We propose and experimentally demonstrate a method for sampling frequency offset (SFO) estimation in optical communication systems based on periodically inserted identical binary prefix. Different from conventional cyclic prefix, binary prefix provides not only high tolerance to chromatic dispersion in dispersive fiber transmission, but also the ability to estimate SFO by simple receiver-side digital signal processing. Moreover, this binary prefix based scheme is generally applicable to any advanced modulation formats.

Self-organization in Balanced State Networks by STDP and Homeostatic Plasticity.

Structural inhomogeneities in synaptic efficacies have a strong impact on population response dynamics of cortical networks and are believed to play an important role in their functioning. However, little is known about how such inhomogeneities could evolve by means of synaptic plasticity. Here we present an adaptive model of a balanced neuronal network that combines two different types of plasticity, STDP and synaptic scaling.

Power-efficient method for IM-DD optical transmission of multiple OFDM signals.

We propose a power-efficient method for transmitting multiple frequency-division multiplexed (FDM) orthogonal frequency-division multiplexing (OFDM) signals in intensity-modulation direct-detection (IM-DD) optical systems. This method is based on quadratic soft clipping in combination with odd-only channel mapping. We show, both analytically and experimentally, that the proposed approach is capable of improving the power efficiency by about 3 dB as compared to conventional FDM OFDM signals under practical bias conditions, making it a viable solution in applications such as optical fiber-wireless integrated systems where both IM-DD optical transmission and OFDM signaling are important.

Demonstration of end-to-end cloud-DSL with a PON-based fronthaul supporting 5.76-Gb/s throughput with 48 eCDMA-encoded 1024-QAM discrete multi-tone signals.

We experimentally demonstrate an end-to-end ultra-broadband cloud-DSL network using passive optical network (PON) based fronthaul with electronic code-division-multiple-access (eCDMA) encoding and decoding. Forty-eight signals that are compliant with the very-high-bit-rate digital subscriber line 2 (VDSL2) standard are transmitted with a record throughput of 5.76 Gb/s over a hybrid link consisting of a 20-km standard single-mode fiber and a 100-m twisted pair.

Simplified flexible-PON upstream transmission using pulse position modulation at ONU and DSP-enabled soft-combining at OLT for adaptive link budgets.

We demonstrate a flexible modulation and detection scheme for upstream transmission in passive optical networks using pulse position modulation at optical network unit, facilitating burst-mode detection with automatic decision threshold tracking, and DSP-enabled soft-combining at optical line terminal. Adaptive receiver sensitivities of -33.1 dBm, -36.

Time-division-multiplexed few-mode passive optical network.

We demonstrate the first few-mode-fiber based passive optical network, effectively utilizing mode multiplexing to eliminate combining loss for upstream traffic. Error-free performance has been achieved for 20-km low-crosstalk 3-mode transmission in a commercial GPON system carrying live Ethernet traffic. The alternative approach of low modal group delay is also analyzed with simulation results over 10 modes.

Flexible TWDM PON system with pluggable optical transceiver modules.

A flexible TWDM PON system is proposed which allows pay-as-you-grow in capacity, supports load balancing among different ODNs, and achieves significant power saving at OLT. Integrated OLT transceiver in enhanced CFP module and low-cost tunable ONU transceiver in SFP+ module are developed, for the first time, for cost effective deployment of TWDM PONs. System experiments demonstrate error free performance with 36 dB power budget in a flexible TWDM PON test bed.

Ionic liquids as recycling solvents for the synthesis of magnetic nanoparticles.

This work describes an easy synthesis (one pot) of MFe(2)O(4) (M = Co, Fe, Mn, and Ni) magnetic nanoparticles MNPs by the thermal decomposition of Fe(Acac)(3)/M(Acac)(2) by using BMI·NTf(2) (1-n-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide) or BMI·PF(6) (1-n-butyl-3-methylimidazolium hexafluorophosphate) ionic liquids (ILs) as recycling solvents and oleylamine as the reducing and surface modifier agent. The effects of reaction temperature and reaction time on the features of the magnetic nanomaterials (size and magnetic properties) were investigated. The growth of the MNPs is easily controlled in the IL by adjusting the reaction temperature and time, as inferred from Fe(3)O(4) MNPs obtained at 150 °C, 200 °C and 250 °C with mean diameters of 8, 10 and 15 nm, respectively.

Preparation and characterization of silica aquasols.

Aquasols containing silica nanoparticles with diameters of 75 to 95 nm were obtained directly by hydrolysis of 2 wt.% tetraethoxysilane (TEOS) in water in the presence of a non-ionic surfactant. The reaction was catalyzed by hydrochloric acid, ammonia, or sodium hydroxide.

High-performance carbon nanotube field effect transistors with a thin gate dielectric based on a self-assembled monolayer.

Individual single-walled carbon nanotube (SWCNT) field effect transistors (FETs) with a 2 nm thick silane-based organic self-assembled monolayer (SAM) gate dielectric have been manufactured. The FETs exhibit a unique combination of excellent device performance parameters. In particular, they operate with a gate-source voltage of only -1 V and exhibit good saturation, large transconductance, and small hysteresis ( View Article and Find Full Text PDF

Raman and theoretical study of the solvent effects on the sizable intramolecular charge transfer in the push-pull 5-(dimethylamino)-5'-nitro-2,2'-bithiophene.

In this paper, we analyze the degree of intramolecular charge transfer in a push-pull pi-conjugated system, 5-(dimethylamino)-5'-nitro-2,2'-bithiophene, from changes in frequencies and relative intensities of its strongest Raman scatterings in a bunch of solvents with different polarities. Density functional theory (DFT) was used as a support of the experimental study. Solvent effects on the molecular and electronic structures and on the vibrational properties were estimated by performing B3LYP/6-31G calculations within the framework of the polarized continuum model (PCM) developed by Tomasi.

Preparation of high quality electrical insulator self-assembled monolayers on gold. Experimental investigation of the conduction mechanism through organic thin films.

Self-assembled monolayers (SAMs) form highly ordered, stable dielectrics on conductive surfaces. Being able to attach larger-area contacts in a MIM (metal-insulator-metal) diode, their electrical properties can be determined. In this paper, the electrical conduction through thiolate SAMs of different alkyl chain lengths formed on gold surfaces were studied and discussed.

Inversion of stereoselectivity by applying mutants of the hydroxynitrile lyase from Manihot esculenta.

The influence of Trp128-substituted mutants of the hydroxynitrile lyase from Manihot esculenta (MeHNL) on the stereoselectivity of MeHNL-catalyzed HCN additions to aldehydes with stereogenic centers, which yield the corresponding cyanohydrins, is described. In rac-2-phenylpropionaldehyde (rac-1) reactions, wild-type (wtMeHNL) and all MeHNL Trp128 mutants are highly (S)-selective toward the (R) enantiomer of rac-1; this results exclusively in (2S,3R)-cyanohydrin ((2S,3R)-2) with > or =96 % de. The (S) enantiomer of rac-1, however, only reacts (S)-selectively with wtMeHNL to give (2S,3S)-2 with 80 % de, whereas with Trp128 mutants, (R) selectivity increases with decreasing size of the amino acids exchanged.

Stereoselective synthesis of cis-p-menth-8-ene-1,7-diol, cis-p-menthane-1,7-diol, and cis-p-menthane-1,7,8-triol.

The natural products cis-p-menthane-1,7-diol (cis-IV), cis-p-menth-8-ene-1,7-diol (cis-I) and cis-p-menthane-1,7,8-triol (cis-II) are obtained starting from the corresponding cis-cyanohydrins, cis-2 and cis-7, respectively, by chemical transformation of the cyano into the hydroxymethyl group. The key step of the synthesis is the very high cis-selectivity (> or = 96 %) of the MeHNL-catalyzed HCN addition to 4-alkylcyclohexanones. From 4-isopropylcyclohexanone (1) the cyanohydrin cis-2 and from 4-(1-methylvinyl)cyclohexanone (6) the cyanohydrin cis-7 result almost quantitatively.

Preparation and photoinduced patterning of azidoformate-terminated self-assembled monolayers.

This paper presents a new method for the patterning of self-assembled monolayers (SAMs) using UV light. Azidoformate-terminated SAMs starting from 18-acetoxy-octadecyltrichlorosilane SAMs on silicon, prepared for the first time, are electrophilic and photosensitive, and can be patterned by UV irradiation through a mask. The resulting structured surfaces are still electrophilic and can be reacted with nucleophilic functions, for example, primary amines.

Low-voltage organic transistors with an amorphous molecular gate dielectric.

Organic thin film transistors (TFTs) are of interest for a variety of large-area electronic applications, such as displays, sensors and electronic barcodes. One of the key problems with existing organic TFTs is their large operating voltage, which often exceeds 20 V. This is due to poor capacitive coupling through relatively thick gate dielectric layers: these dielectrics are usually either inorganic oxides or nitrides, or insulating polymers, and are often thicker than 100 nm to minimize gate leakage currents.

Substrate specificity of mutants of the hydroxynitrile lyase from Manihot esculenta.

Several tryptophan128-substituted mutants of the hydroxynitrile lyase from Manihot esculenta (MeHNL) are constructed and applied in the MeHNL-catalyzed addition of HCN to various aromatic and aliphatic aldehydes as well as to methyl and ethyl ketones to yield the corresponding cyanohydrins. The mutants (especially MeHNL-W128A) are in most cases superior to the wild-type (wt) enzyme when diisopropyl ether is used as the solvent. Substitution of tryptophan128 by an alanine residue enlarges the entrance channel to the active site of MeHNL and thus facilitates access of sterically demanding substrates to the active site, as clearly demonstrated for aromatic aldehydes, especially 3-phenoxybenzaldehyde.