A deep learning ensemble for function prediction of hypothetical proteins from pathogenic bacterial species

Protein function prediction is a crucial task in the post-genomics era due to their diverse irreplaceable roles in a biological system. Traditional methods involved cost-intensive and time-consuming molecular biology techniques but they proved to be ineffective after the outburst of sequencing data through the advent of cost-effective and advanced sequencing techniques. To manage the pace of annotation with that of data generation, there is a shift to computational approaches which are based on homology, sequence and structure-based features, protein-protein interaction networks, phylogenetic profiles, and physicochemical properties, etc. A combination of these features has proven to be promising for protein function prediction in terms of improving prediction accuracy. In the present work, we have employed a combination of features based on sequence, physicochemical property, subsequence and annotation features with a total of 9890 features extracted and/or calculated for 171,212 reviewed prokaryotic proteins of 9 bacterial phyla from UniProtKB, to train a supervised deep learning ensemble model with the aim to categorize a bacterial hypothetical/unreviewed protein’s function into 1739 GO terms as functional classes. The proposed system being fully dedicated to bacterial organisms is a novel attempt amongst various existing machine learning based protein function prediction systems based on mixed organisms. Experimental results demonstrate the success of the proposed deep learning ensemble model based on deep neural network method with F1 measure of 0.7912 on the prepared Test dataset 1 of reviewed proteins.     

Reliable computing with unreliable components: Using separable environments to stabilize long-term information storage

How, in the face of both intrinsic and extrinsic volatility, can unconventional computing fabrics store information over arbitrarily long periods? Here, we argue that the predictable structure of many realistic environments, both natural and artificial, can be used to maintain useful categorical boundaries even when the computational fabric itself is inherently volatile and the inputs and outputs are partially stochastic. As a concrete example, we consider the storage of binary classifications in connectionist networks, although the underlying principles should be applicable to other unconventional computing paradigms. Specifically, we demonstrate that an unsupervised, activity dependent plasticity rule, AHAH (Anti-Hebbian-And-Hebbian), allows binary classifications to remain stable even when the underlying synaptic weights are subject to random noise. When embedded in environments composed of separable features, the weight vector is restricted by the AHAH rule to local attractors representing stable partitions of the input space, allowing unsupervised recovery of stored binary classifications following random perturbations that leave the system in the same basin of attraction. We conclude that the stability of long-term memories may depend not so much on the reliability of the underlying substrate, but rather on the reproducible structure of the environment itself, suggesting a new paradigm for reliable computing with unreliable components.     

Analog computation through high-dimensional physical chaotic neuro-dynamics

Conventional von Neumann computers have difficulty in solving complex and ill-posed real-world problems. However, living organisms often face such problems in real life, and must quickly obtain suitable solutions through physical, dynamical, and collective computations involving vast assemblies of neurons. These highly parallel computations through high-dimensional dynamics (computation through dynamics) are completely different from the numerical computations on von Neumann computers (computation through algorithms). In this paper, we explore a novel computational mechanism with high-dimensional physical chaotic neuro-dynamics. We physically constructed two hardware prototypes using analog chaotic-neuron integrated circuits. These systems combine analog computations with chaotic neuro-dynamics and digital computation through algorithms. We used quadratic assignment problems (QAPs) as benchmarks. The first prototype utilizes an analog chaotic neural network with 800-dimensional dynamics. An external algorithm constructs a solution for a QAP using the internal dynamics of the network. In the second system, 300-dimensional analog chaotic neuro-dynamics drive a tabu-search algorithm. We demonstrate experimentally that both systems efficiently solve QAPs through physical chaotic dynamics. We also qualitatively analyze the underlying mechanism of the highly parallel and collective analog computations by observing global and local dynamics. Furthermore, we introduce spatial and temporal mutual information to quantitatively evaluate the system dynamics. The experimental results confirm the validity and efficiency of the proposed computational paradigm with the physical analog chaotic neuro-dynamics.     

Modeling Topology and Nonlinear Dynamical Behavior of the Weighted Scale-Free Networks

An improved weighted scale-free network, which has two evolution mechanisms： topological growth and strength dynamics, has been introduced. The topology structure of the model will be explored in details in this work. The evolution driven mechanism of Olami-Feder Christensen （OFC） model is added to our model to study the self-organlzed criticality and the dynamical behavior. We also.consider attack mechanism and the study of the model with attack is also investigated in this paper. We tlnd there are differences between the model with attack and without attack.     

Adaptive fuzzy neural network control on the acoustic field in a duct

Base on the principle of the superposition of waves, active noise control is achieved by adaptively tuning a secondary source which produces an anti-noise of equal amplitude and opposite phase with primary source. This paper presents the study on the acoustic attenuation in a duct by using the combination of fuzzy neural network with error back propagation algorithm to control secondary source. The most important advantage of fuzzy inference system is that the structured knowledge is represented in the form of fuzzy IF-THEN rules. But it lacks the ability to accommodate the change of external environments. Combining neural network with fuzzy system can help in this tuning process by adapting fuzzy sets and creating fuzzy rules. The performance of attenuation and control error can be measured by the microphone placed in the downstream of duct. The results of this study, show that the acoustic attenuation by 40 dB for pure-tone noise and nearly 30 dB for dual-tones noise are obtained.     

单块晶体集成的N×N纵横开关网络及其控制算法

设计了一种单块晶体集成的n×n纵横开关(Crossbar)网络。通过综合考虑晶体的双折射和全内双反射现象,以及晶体的电光效应,将构成n×n纵横开关网络的所有单元器件都集成到一块具有电光效应的双折射晶体上。同时,给出了该网络的控制算法,通过对开关工作状态的控制,可以实现任意输入输出通道之间的无阻塞连接。这种单块晶体集成的纵横开关网络具有能量损耗低、无阻塞、易安装、抗干扰等优点,适合于全光网络发展的需要。     

面向下一代航电系统的数据块填装性能

结合光分组交换(OPS)网络和光纤通道(FC)技术的优势, 提出一种下一代航电系统组网方案——基于光分组交换网络的光纤通道技术(FC over OPS)。建立了数学仿真模型, 研究了一种数据块填装算法与网络实时性之间的关系, 分析不同的参数如发送带宽、最低效率门限、发送定时的选取对网络实时性的影响。进一步完成硬件原型设计和仿真, 比较了软件仿真与硬件仿真的结果, 并分析该数据块填装算法的性能。     

Routing strategies in traffic network and phase transition in network traffic flow

The dynamics of information traffic over scale-free networks has been investigated systematically. A series of routing strategies of data packets have been proposed, including the local routing strategy, the next-nearest-neighbour routing strategy, and the mixed routing strategy based on local static and dynamic information. The capacity of the network can be quantified by the phase transition from free flow state to congestion state. The optimal parameter values of each model leading to the highest efficiency of scale-free networked traffic systems have been found. Moreover, we have found hysteretic loop in networked traffic systems with finite packets delivering ability. Such hysteretic loop indicates the existence of the bi-stable state in the traffic dynamics over scale-free networks.      

Synergetic behavior in the cascading failure propagation of scale-free coupled map lattices

In this paper, the whole dynamical process of cascading failures in a class of scale-free coupled map lattices (CML’s), from the occurrence of attack to the end of failure propagation, is investigated. A dynamical model of cascading failures, based on synergetic theory, is constructed. Numerical simulations show that the macroscopic properties of the scale-free CML’s during cascading failure propagation are governed by the general laws of synergetics. This result will be useful in furthering the studies of the prediction and prevention of cascading events in many real-life complex networks.     

Effects of degree correlation on the controllability of networks

In this paper, we investigate the effects of degree correlation on the controllability of undirected networks with different degree-mixing patterns based on simulation analysis. Considering random pinning, max-degree pinning and mix-degree pinning, some relevance factors on controllability such as the control gain and the number of pinned nodes are discussed in detail. It is found that disassortative mixing enhances the network controllability contrast to assortative mixing, to which the network controllability is sensible. Furthermore, too large or too low value of the feedback gain can reduce the controllability. From the viewpoint of practical application, mix-degree pinning strategy is suggested in assortative network, while in disassortative network, max-degree pinning scheme is better than random pinning.