Explore the vast range of titles available.

Low-light image enhancement is challenging in that it needs to consider not only brightness recovery but also complex issues like color distortion and noise, which usually hide in the dark. Simply adjusting the brightness of a low-light image will inevitably amplify those artifacts. To address this difficult problem, this paper proposes a novel end-to-end attention-guided method based on multi-branch convolutional neural network. To this end, we first construct a synthetic dataset with carefully designed low-light simulation strategies. The dataset is much larger and more diverse than existing ones. With the new dataset for training, our method learns two attention maps to guide the brightness enhancement and denoising tasks respectively. The first attention map distinguishes underexposed regions from well lit regions, and the second attention map distinguishes noises from real textures. With their guidance, the proposed multi-branch decomposition-and-fusion enhancement network works in an input adaptive way. Moreover, a reinforcement-net further enhances color and contrast of the output image. Extensive experiments on multiple datasets demonstrate that our method can produce high fidelity enhancement results for low-light images and outperforms the current state-of-the-art methods by a large margin both quantitatively and visually.

Noise reduction is crucial for magnetic anomaly signal detection of underwater ferromagnetic target. Modified wavelet decomposition and reconstruction algorithm is proposed to suppress the colored noise and improve the signal to noise ratio. Hamming window is employed to make the preprocessed signal continuous. Evaluation index based on signal-to-noise ratio function is selected, wavelet decomposition and reconstruction algorithm iterates adaptively to select the optimal order of decomposition and reconstruction. The experiment result emphasizes that the signal-to-noise ratio of novel algorithm is 71.6dB. In this particle, we provide a new method to improve signal-to-noise ratio and enhance real-time signal processing for underwater target magnetic anomaly signal detection.

Inverse stochastic resonance (ISR) is the phenomenon of the response of neuron to noise, which is opposite to the conventional stochastic resonance. In this paper, the ISR phenomena induced by Gaussian and non-Gaussian colored noises are studied in the cases of single Hodgkin–Huxley (HH) neuron and HH neural network, respectively. It is found that the mean firing rate of electrical activities depends on the Gaussian or non-Gaussian colored noises which can induce the phenomenon of ISR. The ISR phenomenon induced by Gaussian colored noise is most obvious under the conditions of low external current, low reciprocal correlation rate and low noise level. The ISR in neural network is more pronounced and lasts longer than the duration of a single neuron. However, the ISR phenomenon induced by non-Gaussian colored noise is apparent under low noise correlation time or low departure from Gaussian noise, and the ISR phenomena show different duration ranges under different parameter values. Furthermore, the transition of mean firing rate is more gradual, the ISR lasts longer, and the ISR phenomenon is more pronounced under the non-Gaussian colored noise. The ISR is a common phenomenon in neurodynamics; our results might provide novel insights into the ISR phenomena observed in biological experiments.

Reinforcement learning involves decision-making in dynamic and uncertain environments and constitutes a crucial element of artificial intelligence. In our previous work, we experimentally demonstrated that the ultrafast chaotic oscillatory dynamics of lasers can be used to efficiently solve the two-armed bandit problem, which requires decision-making concerning a class of difficult trade-offs called the exploration–exploitation dilemma. However, only two selections were employed in that research; hence, the scalability of the laser-chaos-based reinforcement learning should be clarified. In this study, we demonstrated a scalable, pipelined principle of resolving the multi-armed bandit problem by introducing time-division multiplexing of chaotically oscillated ultrafast time series. The experimental demonstrations in which bandit problems with up to 64 arms were successfully solved are presented where laser chaos time series significantly outperforms quasiperiodic signals, computer-generated pseudorandom numbers, and coloured noise. Detailed analyses are also provided that include performance comparisons among laser chaos signals generated in different physical conditions, which coincide with the diffusivity inherent in the time series. This study paves the way for ultrafast reinforcement learning by taking advantage of the ultrahigh bandwidths of light wave and practical enabling technologies.

In this paper, the stochastic bifurcations and the performance analysis of a strongly nonlinear tri-stable energy harvesting system with colored noise are investigated. Using the stochastic averaging method, the averaged Fokker–Plank–Kolmogorov equation and the stationary probability density (SPD) of the amplitude are obtained, respectively. Meanwhile, the Monte Carlo simulations are performed to verify the effectiveness of the theoretical results. D-bifurcation is studied through the largest Lyapunov exponent calculations, which implies the system undergoes D-bifurcation twice with increasing the nonlinear stiffness coefficients. The effects of the nonlinear stiffness coefficients, noise intensity and correlation time on P-bifurcation are discussed by the qualitative changes of the SPD. Moreover, the relationship between D- and P-bifurcation is explored. If the strength of stochastic jump has obvious gap with respect to the two statuses before and after the occurrence of P-bifurcation, the D-bifurcation will happen, too. Finally, the performance and the capability of harvesting energy from ambient random excitation are analyzed.

We study a system of interacting active particles, propelled by colored noises, characterized by an activity time τ, and confined by a single-well anharmonic potential. We assume pair-wise repulsive forces among particles, modelling the steric interactions among microswimmers. This system has been experimentally studied in the case of a dilute suspension of Janus particles confined through acoustic traps. We observe that already in the dilute regime - when inter-particle interactions are negligible - increasing the persistent time, τ , pushes the particles away from the potential minimum, until a saturation distance is reached. We compute the phase diagram (activity versus interaction length), showing that the interaction does not suppress this delocalization phenomenon but induces a liquid- or solid-like structure in the densest regions. Interestingly a reentrant behavior is observed: a first increase of τ from small values acts as an effective warming, favouring fluidization; at higher values, when the delocalization occurs, a further increase of τ induces freezing inside the densest regions. An approximate analytical scheme gives fair predictions for the density profiles in the weakly interacting case. The analysis of non-equilibrium heat fluxes reveals that in the region of largest particle concentration equilibrium is restored in several aspects.

Tipping in multistable systems occurs usually by varying the input slightly, resulting in the output switching to an often unsatisfactory state. This phenomenon is manifested in thermoacoustic systems. The thermoacoustic instability may lead to the disintegration of rocket engines, gas turbines and aeroengines, so it is necessary to design control measures for its suppression. It was speculated that such unwanted instability states may be dodged by changing the bifurcation parameters quickly enough, and compared with the white noise discussed in [ 1 ], colored noise with nonzero correlation time is more practical and important to the system. Thus, in this work, based on a fundamental mathematical model of thermoacoustic systems driven by colored noise, the corresponding Fokker–Planck–Kolmogorov equation of the amplitude is derived by using a stochastic averaging method. A transient dynamical behavior is identified through a probability density analysis. We find that both a relatively higher rate of change of parameters and change in the correlation time of the noise are beneficial to dodge thermoacoustic instability, while a relatively large noise intensity is a disadvantageous factor. More interestingly and importantly, power-law relationships between the maximum amplitude and the noise parameters are uncovered, and the probability of successfully dodging a thermoacoustic instability is calculated. These results serve as a guidance for the design of engines and to propose an effective control strategy, which is of great significance to aerospace-related fields.

Chatter is a kind of unstable vibration in high-speed milling process, leading to poor surface quality of workpiece, significant tool wear, and severe noise. In order to avoid these negative effects of milling chatter, the detection of chatter at early stage is highly needed. In this paper, an early-stage chatter detection method based on variational mode decomposition (VMD) and difference of power spectral entropy (ΔPSE) is presented. Considering that the existence of possible colored noise in the monitoring signals, which might lead to the misjudgment of chatter detection, the signals monitored at spindle’s idling is utilized to identify these noise components. In order to separate the needed chatter-sensitive sub-signals, VMD is utilized to decompose the original signals into a series of intrinsic mode functions (IMFs), and the chatter-sensitive sub-signals are obtained by adding the IMFs whose central frequencies are closed to the milling system’s natural frequency. After that, an adaptive filter is utilized to filter out the harmonics of spindle-speed frequency and the identified colored noise components. Then, a dimensionless indicator is designed, which is determined as the difference of power spectral entropy (ΔPSE) of signals without and with filtering. A series of experiments are also performed, and the results indicate that the presented methodology can detect the chatter at early stage and is applicable in different cutting conditions, which is very important in the practical application.

Instrumental variables are widely used to identify linear dynamical systems. The advantages of instrumental variables include low computational complexity, as well as the possibility of identification for different models of color noise. Often the method of instrumental variables leads to ill-conditioned problems, which significantly limits the application of this method. The paper proposes a solution to the problem of extended instrumental variables based on augmented normal equations. Test examples showed a high accuracy of the proposed approach.

Exploration of Underwater Images will be a challenging task due to its degradations by haze, blur, colour cast and noise. In order to extract the latent information from the blurred portion, its parameters are estimated and various algorithms were analysed. Initially, length and angle of motion blurred image are estimated. Then, filters and restoration algorithms were implemented for controlling of ringing artifacts, removal of noise and latent information restoration. Some of the model based classical methods were implemented and analysed for underwater motion blurred image restoration.