Food habits of Sowerby's beaked whales (Mesoplodon bidens) taken in the pelagic drift gillnet fishery of the western North Atlantic

We describe the food habits of the Sowerby’s beaked whale (Mesoplodon bidens) from observations of 10 individuals taken as bycatch in the pelagic drift gillnet fishery for Swordfish (Xiphias gladius) in the western North Atlantic and 1 stranded individual from Kennebunk, Maine. The stomachs of 8 bycaught whales were intact and contained prey. The diet of these 8 whales was dominated by meso- and benthopelagic fishes that composed 98.5% of the prey items found in their stomachs and cephalopods that accounted for only 1.5% of the number of prey. Otoliths and jaws representing at least 31 fish taxa from 15 families were present in the stomach contents. Fishes, primarily from the families Moridae (37.9% of prey), Myctophidae (22.9%), Macrouridae (11.2%), and Phycidae (7.2%), were present in all 8 stomachs. Most prey were from 5 fish taxa: Shortbeard Codling (Laemonema barbatulum) accounted for 35.3% of otoliths, Cocco’s Lanternfish (Lobianchia gemellarii) contributed 12.9%, Marlin-spike (Nezumia bairdii) composed 10.8%, lanternfishes (Lampanyctus spp.) accounted for 8.4%; and Longfin Hake (Phycis chesteri) contributed 6.7%. The mean number of otoliths per stomach was 1196 (range: 327–3452). Most of the fish prey found in the stomachs was quite small, ranging in length from 4.0 to 27.7 cm. We conclude that the Sowerby’s beaked whales that we examined in this study fed on large numbers of relatively small meso- and benthopelagic fishes that are abundant along the slope and shelf break of the western North Atlantic.

Bayesian and L1 Approaches to Sparse Unsupervised Learning

The use of L1 regularisation for sparse learning has generated immense research interest, with successful application in such diverse areas as signal acquisition, image coding, genomics and collaborative filtering. While existing work highlights the many advantages of L1 methods, in this paper we find that L1 regularisation often dramatically underperforms in terms of predictive performance when compared with other methods for inferring sparsity. We focus on unsupervised latent variable models, and develop L1 minimising factor models, Bayesian variants of "L1", and Bayesian models with a stronger L0-like sparsity induced through spike-and-slab distributions. These spike-and-slab Bayesian factor models encourage sparsity while accounting for uncertainty in a principled manner and avoiding unnecessary shrinkage of non-zero values. We demonstrate on a number of data sets that in practice spike-and-slab Bayesian methods outperform L1 minimisation, even on a computational budget. We thus highlight the need to re-assess the wide use of L1 methods in sparsity-reliant applications, particularly when we care about generalising to previously unseen data, and provide an alternative that, over many varying conditions, provides improved generalisation performance.

Millisecond processing beyond chip technology: From electronics to photonics

There is a clear and increasing interest in short time annealing processing far below one second, i.e. the lower limit of Rapid Thermal Processing (RTP) called spike annealing. This was driven by the need of suppressing the so-called Transient Enhanced Diffusion in advanced boronimplanted shallow pn-junctions in silicon technology. Meanwhile the interest in flash lamp annealing (FLA) in the millisecond range spread out into other fields related to silicon technology and beyond. This paper reports on recent experiments regarding shallow junction engineering in germanium, annealing of ITO layers on glass and plastic foil to form an conductive layer as well as investigations which we did during the last years in the field of wide band gap semiconductor materials (SiC, ZnO). A more common feature evolving from our work was related to the modeling of wafer stress during millisecond thermal processing with flash lamps. Finally recent achievements in the field of silicon-based light emission basing on Metal-Oxide-Semiconductor Light Emitting Devices will be reported. © 2007 IEEE.

Nonparametric mixtures of factor analyzers

The mixtures of factor analyzers (MFA) model allows data to be modeled as a mixture of Gaussians with a reduced parametrization. We present the formulation of a nonparametric form of the MFA model, the Dirichlet process MFA (DPMFA). The proposed model can be used for density estimation or clustering of high dimensiona data. We utilize the DPMFA for clustering the action potentials of different neurons from extracellular recordings, a problem known as spike sorting. DPMFA model is compared to Dirichlet process mixtures of Gaussians model (DPGMM) which has a higher computational complexity. We show that DPMFA has similar modeling performance in lower dimensions when compared to DPGMM, and is able to work in higher dimensions. ©2009 IEEE.

Stability enhancement by casing grooves: The importance of stall inception mechanism and solidity

This paper concerns the optimisation of casing grooves and the important influence of stall inception mechanism on groove performance. Installing casing grooves is a well known technique for improving the stable operating range of a compressor, but the wide-spread use of grooves is restricted by the loss of efficiency and flow capacity. In this paper, laboratory tests are used to examine the conditions under which casing treatment can be used to greatest effect. The use of a single casing groove was investigated in a recently published companion paper. The current work extends this to multiple-groove treatments and considers their performance in relation to stall inception mechanisms. Here it is shown that the stall margin gain from multiple grooves is less than the sum of the gains if the grooves were used individually. By contrast, the loss of efficiency is additive as the number of grooves increases. It is then shown that casing grooves give the greatest stall margin improvement when used in a compressor which exhibits spike-type stall inception, while modal activity before stall can dramatically reduce the effectiveness of the grooves. This finding highlights the importance of being able to predict the stall inception mechanism which might occur in a given compressor before and after grooves are added. Some published prediction techniques are therefore examined, but found wanting. Lastly, it is shown that casing grooves can, in some cases, be used to remove rotor blades and produce a more efficient, stable and light-weight rotor. © 2010 by ASME.

A neuro-inspired method for data rate limited feedback control

The nervous system implements a networked control system in which the plants take the form of limbs, the controller is the brain, and neurons form the communication channels. Unlike standard networked control architectures, there is no periodic sampling, and the fundamental units of communication contain little numerical information. This paper describes a novel communication channel, modeled after spiking neurons, in which the transmitter integrates an input signal and sends out a spike when the integral reaches a threshold value. The reciever then filters the sequence of spikes to approximately reconstruct the input signal. It is shown that for appropriate choices of channel parameters, stable feedback control over these spiking channels is possible. Furthermore, good tracking performance can be achieved. The data rate of the channel increases linearly with the size of the inputs. Thus, when placed in a feedback loop, small loop gains imply a low data rate. ©2010 IEEE.

Reinforcement learning using a continuous time actor-critic framework with spiking neurons.

Animals repeat rewarded behaviors, but the physiological basis of reward-based learning has only been partially elucidated. On one hand, experimental evidence shows that the neuromodulator dopamine carries information about rewards and affects synaptic plasticity. On the other hand, the theory of reinforcement learning provides a framework for reward-based learning. Recent models of reward-modulated spike-timing-dependent plasticity have made first steps towards bridging the gap between the two approaches, but faced two problems. First, reinforcement learning is typically formulated in a discrete framework, ill-adapted to the description of natural situations. Second, biologically plausible models of reward-modulated spike-timing-dependent plasticity require precise calculation of the reward prediction error, yet it remains to be shown how this can be computed by neurons. Here we propose a solution to these problems by extending the continuous temporal difference (TD) learning of Doya (2000) to the case of spiking neurons in an actor-critic network operating in continuous time, and with continuous state and action representations. In our model, the critic learns to predict expected future rewards in real time. Its activity, together with actual rewards, conditions the delivery of a neuromodulatory TD signal to itself and to the actor, which is responsible for action choice. In simulations, we show that such an architecture can solve a Morris water-maze-like navigation task, in a number of trials consistent with reported animal performance. We also use our model to solve the acrobot and the cartpole problems, two complex motor control tasks. Our model provides a plausible way of computing reward prediction error in the brain. Moreover, the analytically derived learning rule is consistent with experimental evidence for dopamine-modulated spike-timing-dependent plasticity.

Changing the responses of cortical neurons from sub- to suprathreshold using single spikes in vivo.

Action Potential (APs) patterns of sensory cortex neurons encode a variety of stimulus features, but how can a neuron change the feature to which it responds? Here, we show that in vivo a spike-timing-dependent plasticity (STDP) protocol-consisting of pairing a postsynaptic AP with visually driven presynaptic inputs-modifies a neurons' AP-response in a bidirectional way that depends on the relative AP-timing during pairing. Whereas postsynaptic APs repeatedly following presynaptic activation can convert subthreshold into suprathreshold responses, APs repeatedly preceding presynaptic activation reduce AP responses to visual stimulation. These changes were paralleled by restructuring of the neurons response to surround stimulus locations and membrane-potential time-course. Computational simulations could reproduce the observed subthreshold voltage changes only when presynaptic temporal jitter was included. Together this shows that STDP rules can modify output patterns of sensory neurons and the timing of single-APs plays a crucial role in sensory coding and plasticity.DOI:http://dx.doi.org/10.7554/eLife.00012.001.

Code-specific policy gradient rules for spiking neurons

Although it is widely believed that reinforcement learning is a suitable tool for describing behavioral learning, the mechanisms by which it can be implemented in networks of spiking neurons are not fully understood. Here, we show that different learning rules emerge from a policy gradient approach depending on which features of the spike trains are assumed to influence the reward signals, i.e., depending on which neural code is in effect. We use the framework of Williams (1992) to derive learning rules for arbitrary neural codes. For illustration, we present policy-gradient rules for three different example codes - a spike count code, a spike timing code and the most general "full spike train" code - and test them on simple model problems. In addition to classical synaptic learning, we derive learning rules for intrinsic parameters that control the excitability of the neuron. The spike count learning rule has structural similarities with established Bienenstock-Cooper-Munro rules. If the distribution of the relevant spike train features belongs to the natural exponential family, the learning rules have a characteristic shape that raises interesting prediction problems.

Determinations of residual furazolidone and its metabolite, 3-amino-2-oxazolidinone (AOZ), in fish feeds by HPLC-UV and LC-MS/MS, respectively

The antibacterial drug furazolidone belonging to the group of nitrofuran antibacterial agents has been widely used as an antibacterial and antiprotozoal feed additive for poultry, cattle, and farmed fish in China. During application a large proportion of the administered drug may reach the environment directly or via feces. Although the use of furazolidone is prohibited in numerous countries, there are indications of its illegal use. It is known that furazolidone can be rapidly metabolized to 3-amino-2-oxazolidinone (AOZ) in the body of the target organism. In this study, a total of 21 fish feed samples, including 17 commercial fish feeds from local markets in China (representing 15 different formulations) and 4 fish feeds obtained from Germany and Turkey, respectively, are analyzed to determine whether the drug is still illegally used or commercially available feeds are contaminated by this drug. High-performance liquid chromatography (HPLC) and liquid chromatography-electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) methods have been implemented to determine furazolidone and its metabolite AOZ in fish feeds containing animal protein, respectively. An efficient and convenient cleanup method for the determination of furazolidone in fish feeds is developed, and a simple cleanup method for the determination of AOZ is used. Method recoveries for samples used were determined as 87.7-98.3% for furazolidone at two spike levels of 2.0 and 5.0 ng g(-1) and as 95.6-102.8% for AOZ at spike levels of 0.4 and 0.8 ng g(-1). Limits of detections were 0.4 ng g(-1) for furazolidone and 0.05 ng g(-1) for AOZ. The established methods are therefore suitable for the determination of furazolidone and its metabolite AOZ in fish feeds at trace contamination levels. Using the established methods, all fish feed samples have been proved to be furazolidone negative; however, AOZ is tested in 16 of 17 fish feeds obtained from local markets in the Hubei province of China, with a positive rate as high as 94.1%.

Multi-channel micro neural probe fabricated with SOI

Silicon-on-insulator (SOI) substrate is widely used in micro-electro-mechanical systems (MEMS). With the buried oxide layer of SOI acting as an etching stop, silicon based micro neural probe can be fabricated with improved uniformity and manufacturability. A seven-record-site neural probe was formed by inductive-coupled plasma (ICP) dry etching of an SOI substrate. The thickness of the probe is 15 mu m. The shaft of the probe has dimensions of 3 mmx100 mu mx15 mu m with typical area of the record site of 78.5 mu m(2). The impedance of the record site was measured in-vitro. The typical impedance characteristics of the record sites are around 2 M Omega at 1 kHz. The performance of the neural probe in-vivo was tested on anesthetic rat. The recorded neural spike was typically around 140 mu V. Spike from individual site could exceed 700 mu V. The average signal noise ratio was 7 or more.

Characteristics of oxide-free InGaAlAs layers grown by narrow stripe selective MOVPE

Narrow stripe selective MOVPE has been used to grow high quality oxide-free InGaAlAs layers on an InP substrate patterned with SiO2 masks at optimized growth conditions. Mirror-like surface morphologies and abrupt cross sections are obtained in all samples without spike growth at the mask edge. For the narrow stripe selectively grown InGaAlAs layers with a mesa width of about 1.2 mu m, a bandgap wavelength shift of 70 nm, a photoluminescence (PL) intensity of more than 80% and a PL full width at half maximum (FWHM) of less than 60 meV are obtained simultaneously with a small mask width variation from 0 to 40 mu m. The characteristics of the thickness enhancement ratio and the PL spectrum dependence on the mask width are presented and explained by considering both the migration effect from a masked region and the lateral vapour diffusion effect.

Multi channel micro neural probe fabricated with SOI

Extracellular neural recording requires neural probes having more recording sites as well as limited volumes. With its mechanical characteristic and abundant process method, Silicon is a kind of material fit for producing neural probe. Silicon on insulator (SOI) is adopted in this paper to fabricate neural probes. The uniformity and manufacturability are improved. The fabricating process and testing results of a series of Multi channel micro neural probes were reported. The thickness of the probe is 15 mu m-30 mu m. The typical impedance characteristics of the record sites are around 2M Omega at 1k Hz. The performance of the neural probe in-vivo was tested on anesthetic rat. The recorded neural spike was typically around 140 mu V. Spike recorded from individual site could exceed 700 mu V. The average signal noise ratio was 7 or more.

on string languages generated by spiking neural p systems

We continue the study of spiking neural P systems by considering these computing devices as binary string generators: the set of spike trains of halting computations of a given system constitutes the language generated by that system. Although the "direct" generative capacity of spiking neural P systems is rather restricted (some very simple languages cannot be generated in this framework), regular languages are inverse-morphic images of languages of finite spiking neural P systems, and recursively enumerable languages are projections of inverse-morphic images of languages generated by spiking neural P systems.

Selective Area Growth InGaAsP by MOVPE

The wide stripe (ISjum) selective area growth (SAG) of InGaAsP by low pressure MOVPE is systematically investigated. The characteristics of the growth ratios,thickness enhancement factors .bandgap modulation,and composition modulation vary with the growth conditions such as mask width,growth pressure. Flux of III-group precursors are outlined and the rational mechanism behind SAG MOVPE is explained. In addition,the surface spike of the SAG InGaAsP is shown and the course of it is given by the variation of V /III .