Search

Understanding how the brain solves demanding computational tasks is one of the most exciting scientific challenges of our times. So far, recurrently connected artificial neural network models (RANNs) were primarily used to reverse-engineer brain computations. We show that it is now also feasible to reverse-engineer computations of detailed data-based large-scale models of cortical microcircuits. Furthermore, results of these analyses produce hypotheses that can readily be tested in biological experiments since they clarify from which neurons one needs to record and what type of information can be expected at specific time points during a trial. We apply this approach to a demanding visual processing task that has often been used in mouse experiments. Both the cortical microcircuit model and RANNs can solve this task as well as the mouse. But the resulting network dynamics matches only for the cortical microcircuit model experimental data on the sparseness of network activity and the impact of individual neurons on the network decision. Reverse-engineering of the computation in the cortical microcircuit model suggests that a particular subset of neurons causes a bifurcation of the network dynamics that triggers the network decision. Altogether, our results introduce a new type of neural network model for brain computations.Teaser: Large-scale modeling on supercomputers provides new tools for understanding how the brain computes

We analyze visual processing capabilities of a large-scale model for area V1 that arguably provides the most comprehensive accumulation of anatomical and neurophysiological data to date. We find that this brain-like neural network model can reproduce a number of characteristic visual processing capabilities of the brain, in particular the capability to solve diverse visual processing tasks, also on temporally dispersed visual information, with remarkable robustness to noise. This V1 model, whose architecture and neurons markedly differ from those of deep neural networks used in current artificial intelligence (AI), such as convolutional neural networks (CNNs), also reproduces a number of characteristic neural coding properties of the brain, which provides explanations for its superior noise robustness. Because visual processing is substantially more energy efficient in the brain compared with CNNs in AI, such brain-like neural networks are likely to have an impact on future technology: as blueprints for visual processing in more energy-efficient neuromorphic hardware.

Recent evidence has demonstrated that during visual spatial attention sampling, neural activity and behavioral performance exhibit large fluctuations. To understand the origin of these fluctuations and their functional role, here, we introduce a mechanism based on the dynamical activity pattern (attention spotlight) emerging from neural circuit models in the transition regime between different dynamical states. This attention activity pattern with rich spatiotemporal dynamics flexibly samples from different stimulus locations, explaining many key aspects of temporal fluctuations such as variable theta oscillations of visual spatial attention. Moreover, the mechanism expands our understanding of how visual attention exploits spatially complex fluctuations characterized by superdiffusive motion in space and makes experimentally testable predictions. We further illustrate that attention sampling based on such spatiotemporal fluctuations provides profound functional advantages such as adaptive switching between exploitation and exploration activities and is particularly efficient at sampling natural scenes with multiple salient objects.

Fluorinated carbon (CFx), a thriving member of the carbonaceous derivative, possesses various excellent properties of chemically stable, tunable bandgap, good thermal conductivity and stability, and super-hydrophobic due to its unique structures and polar C-F bonding. Herein, we present a brief review of the recent development of fluorinated carbon materials in terms of structures, properties and preparation techniques. Meanwhile, the applications in energy conversions and storage devices, biomedicines, gas sensors, electronic devices, and microwave absorption devices are also presented. The fluorinated carbon contains various types of C-F bonds including ionic, semi-ionic and covalent C-F, C-F2, C-F3 bonds with tunable F/C ratios. The controllable designing of C-F bonding and F/C ratios play a key role to optimize the properties of fluorinated carbon materials. Until now, the potential issues and future opportunities of fluorinated carbon are proposed. The present review will provide a direction for tuning C-F bonding and F/C ratios, developing a safe and efficient fluorination method and popularizing the applications of fluorinated carbon materials.

Learning in deep neural networks (DNNs) is implemented through minimizing a highly non-convex loss function, typically by a stochastic gradient descent (SGD) method. This learning process can effectively find good wide minima without being trapped in poor local ones. We present a novel account of how such effective deep learning emerges through the interactions of the SGD and the geometrical structure of the loss landscape. Rather than being a normal diffusion process (i.e. Brownian motion) as often assumed, we find that the SGD exhibits rich, complex dynamics when navigating through the loss landscape; initially, the SGD exhibits anomalous superdiffusion, which attenuates gradually and changes to subdiffusion at long times when the solution is reached. Such learning dynamics happen ubiquitously in different DNNs such as ResNet and VGG-like networks and are insensitive to batch size and learning rate. The anomalous superdiffusion process during the initial learning phase indicates that the motion of SGD along the loss landscape possesses intermittent, big jumps; this non-equilibrium property enables the SGD to escape from sharp local minima. By adapting the methods developed for studying energy landscapes in complex physical systems, we find that such superdiffusive learning dynamics are due to the interactions of the SGD and the fractal-like structure of the loss landscape. We further develop a simple model to demonstrate the mechanistic role of the fractal loss landscape in enabling the SGD to effectively find global minima. Our results thus reveal the effectiveness of deep learning from a novel perspective and have implications for designing efficient deep neural networks., 10 pages, 8 figures, a new angle to unravel the learning dynamics of SGD in DNNs

Two-dimensional graphene is easy to aggregate and hard to transfer because of the van der Waals forces between graphene layers. The three-dimensional (3D) graphite-like capsules (GCs) disperse well and exhibit robust structural stability and have a potential for promising applications in energy storage, drug delivery, catalyst substrate, etc. Here, we develop a catalytic chemical vapor deposition route for the synthesis of 3D GCs with unprecedented thickness and morphology control. The theoretical results indicate that the strong chemisorption and charge transfer between ZnO and acetylene allow the layer-by-layer formation of GCs that is later confirmed by the experimental data. The 3D GCs are synthesized with thickness control from three atomic layers to ∼300 atomic layers over various oxides with predesigned shapes of microscale tetrapods, nanospheres, nanorods, nanocubes, etc. In particular, the formation mechanism of GCs is investigated using the ZnO surface as a catalyst through systematically experi...

A method is designed to improve the luminescence of AlN-based phosphors by tuning the band structure and crystal structure due to alloying with GaN. The pure (Al,Ga)N:Eu phosphors were initially prepared by gas-phase reaction in an NH3 atmosphere. GaN alloying was used to expand the crystal lattice of AlN due to Ga3+ substituting for smaller Al3+ ions, making dissolution of Eu2+ easier. The dissolution of Ga in the AlN lattice was proven by the result of the Rietveld refinement and the increase in lattice parameters with increasing Ga content. To introduce other energy states mixing with the 5d states of Eu2+, Ga doping was also used to tune the band structure of AlN by acting on Eu2+ ions. The theoretical result was analyzed using the Cambridge Sequential Total Energy Package (CASTEP). According to the calculated total and atom resolved partial density of states, it was observed that the Ga 5p states contribute a large portion to the corresponding Eu2+ absorption band in (Al,Ga)N:Eu phosphors. As a consequence, an enhanced emission intensity at 470 nm and a high quantum efficiency for excitation at 330 nm was obtained despite of stronger thermal quenching of the (Al,Ga)N:Eu phosphors compared with AlN:Eu.

Cu nanocrystals are applied extensively in several fields, particularly in the microelectron, sensor, and catalysis. The catalytic behavior of Cu nanocrystals depends mainly on the structure and particle size. In this work, formation of high-purity Cu nanocrystals is studied using a common chemical vapor deposition precursor of cupric tartrate. This process is investigated through a combined experimental and computational approach. The decomposition kinetics is researched via differential scanning calorimetry and thermogravimetric analysis using Flynn-Wall-Ozawa, Kissinger, and Starink methods. The growth was found to be influenced by the factors of reaction temperature, protective gas, and time. And microstructural and thermal characterizations were performed by X-ray diffraction, scanning electron microscopy, transmission electron microscopy, and differential scanning calorimetry. Decomposition of cupric tartrate at different temperatures was simulated by density functional theory calculations under the generalized gradient approximation. High crystalline Cu nanocrystals without floccules were obtained from thermal decomposition of cupric tartrate at 271°C for 8 h under Ar. This general approach paves a way to controllable synthesis of Cu nanocrystals with high purity.

HM Clarke, G Chen. Enhanced peripheral neovascularization of musculocutaneous flaps following pedicle ligation in the pig. Can J Plast Surg 1994:2(4): 159-163. The augmentation of blood flow from the periphery following dominant pedicle ligation was studied in island latissimus dorsi muscle and musculocutaneous flaps in the pig. In the first study, a musculocutaneous flap and a muscle flap covered with a skin graft were raised on two sides of the same pig (n=25). Silicone rubber sheeting inhibited revascularization from the base. From one to seven days later the dominant vascular pedicles were ligated and blood flow from neovascularization to the flap was assessed immediately using radiolabelled 15 μm microspheres. The peripheral neovascularization of muscle and musculocutaneous flaps was not different when assessed at the time of pedicle ligation. Comparison with a previous study of flap survival, however, revealed that increased blood flow was seen in flaps in which the pedicle had been ligated two days prior to rather than at the time of assessment. Therefore, in a second study, musculocutaneous flaps were raised bilaterally and from one to eight days later the pedicle was ligated on one side only (n=29). After two days of further neovascularization the blood flow to the components of the flaps was assessed using 15 μm microspheres. Significant enhancement of peripheral neovascularization was seen in all animals in which two days had elapsed between pedicle ligation and the assessment of neovascularization (P

The facile preparation of ZnO possessing high visible luminescence intensity remains challenging due to an unclear luminescence mechanism. Here, two basic approaches are proposed to enhance the luminescent intensity based on the theoretical analysis over surface defects. Based on the deduction, we introduce a methodology for obtaining hybrid tetrapod-like zinc oxide (T-ZnO), decorated by carbon nanomarterials on T-ZnO surfaces through the catalytic chemical vapor deposition approach. The intensity of the T-ZnO green emission can be modulated by topography and the proportion of carbon. Under proper experiment conditions, the carbon decorating leads to dramatically enhanced luminescence intensity of T-ZnO from 400 to 700 nm compared with no carbon decorated, which elevates this approach to a simple and effective method for the betterment of fluorescent materials in practical applications.

The rational treatment of severe microtia or anotia should be to restore not only the form but also the auditory function of the ear. To achieve this aim it is obvious that plastic surgeons must co-operate closely with an otologist: yet this degree of co-operation is seldom seen. This paper describes a two-stage procedure for total ear reconstruction with auditory function. In stage I, a tympanoplasty is performed before transplantation of the carved cartilage framework. Stage II consists of elevation of the reconstructed auricle and transplantation of a full-thickness skin graft to the post-auricular defect. Certain modifications of the operative techniques that we have found useful are discussed in detail.