Narrow Results

Typically, deep learning models for image segmentation tasks are trained using large datasets of images annotated at the pixel level, which can be expensive and highly time-consuming. A way to reduce the amount of annotated images required for training is to adopt a semi-supervised approach. In this regard, generative deep learning models, concretely Generative Adversarial Networks (GANs), have been adapted to semi-supervised training of segmentation tasks. This work proposes MaskGDM, a deep learning architecture combining some ideas from EditGAN, a GAN that jointly models images and their segmentations, together with a generative diffusion model. With careful integration, we find that using a generative diffusion model can improve EditGAN performance results in multiple segmentation datasets, both multi-class and with binary labels. According to the quantitative results obtained, the proposed model improves multi-class image segmentation when compared to the EditGAN and DatasetGAN models, respectively, by $4.5\%$ and $5.0\%$. Moreover, using the ISIC dataset, our proposal improves the results from other models by up to $11\%$ for the binary image segmentation approach.

In the background of big data era, the ability to accurately forecast the number of the Internet users has considerable implications for evaluating the growing trend of a newly-developed business. In this paper, we use four models, the Gompertz model, the Logistic model, the Bass model, and the Lotka–Volterra model, to forecast the Internet population in China with the historical data during 2007 to 2014. We compare the prediction accuracy of the four models using the criterions such as the mean absolute percentage error (MAPE), the mean absolute error (MAE) and the root mean square error (RMSE). We find that the Lotka–Volterra model has the highest prediction accuracy. Moreover, we use the Lotka–Volterra model to investigate the relationship between the rural Internet users and the urban Internet users in China. The estimation results show that the relationship is commensalism.

The origin of the knee in the energy spectrum of cosmic rays is one of the central questions of high-energy astrophysics. One possible explanation is the energy dependent leakage of nuclei from the Galaxy due to their propagation. The latter is investigated in a combined method using numerical calculations of trajectories and the diffusion approximation. The life time of cosmic rays in the Galaxy and the corresponding pathlength are presented. The resulting energy spectra as observed at Earth are discussed and compared to experimental data.

In this paper, we propose a new algorithm called LL-Diff, which is innovative compared to traditional augmentation methods in that it introduces the sampling method of Langevin dynamics. This sampling approach simulates the motion of particles in complex environments and can better handle noise and details in low-light conditions. We also incorporate a causal attention mechanism to achieve causality and address the issue of confounding effects. This attention mechanism enables us to better capture local information while avoiding over-enhancement. We have conducted experiments on the LOL-V1 and LOL-V2 datasets, and the results show that LL-Diff significantly improves computational speed and several evaluation metrics, demonstrating the superiority and effectiveness of our method for low-light image enhancement tasks. The code will be released on GitHub when the paper has been accepted.

The isospin effect on particle emission for fissioning isobaric sources of ¹¹⁰Tc, ¹¹⁰Pd, ¹¹⁰In and for isotopic sources of ^110,117,124In is explored in the framework of the Smoluchowski equation. A statistical model including dissipation is employed to study particle emission in asymmetric and symmetric fission of the In nucleus with different isospins. Calculations show that for a fissioning nucleus which has a larger isospin, charged particle multiplicities are no longer sensitive to dissipation strength or the fission time scale. Hence, for those systems with very high isospins, protons and α particles cannot be used as probes of the dissipation in the fission of hot nuclei. This conclusion does not depend on the mass asymmetry of the fission process.

In this work, a mathematical model for simulating the thermochemical boronizing process is presented. The diffusion model used in this paper is based on Fick’s laws by solving the mass balance equation of the (FeB/Fe₂B) interface. In this developed model, the effect of boride incubation times during the formation of Fe₂B layers on Armco iron was considered. To demonstrate the validity of our calculations, the simulation results are compared with experimentally obtained data on borided Armco iron, which allowed us to verify the validity of the model. Therefore, a good concordance was observed when comparing the experimental parabolic growth constants taken from the literature with our simulated values of the parabolic growth constants from the present diffusion model. From this study, it has been found that the incubation time has a very important influence on the evolution of the kinetics of the boride layers.

Modeling diffusion dynamics of multi-generation innovation requires a critical examination of external factors that may affect its diffusion process. It has been observed that due to companies continuously varying marketing strategies, the adoption rate of an innovation alters with time. However, there are other factors such as the launch of a new competitive product or improved product generation, which may affect the growth of an innovation. The time-instance at which these changes are observed is called change-point. Motivated by this phenomenon, the present research identifies the launch of a new generation as a change-point where adoption function of the previous generation experiences a structural change. The objective of the current research is to improve the forecasting accuracy of a diffusion model for technological innovations by integrating essential factors that affect the diffusion process. From the findings of empirical analysis, it can be inferred that the proposed two-generational diffusion model illustrates the diffusion pattern of Dynamic Random Access Memory (DRAM) semiconductors remarkably well. In fact, the computed results show that the suggested model has better forecasting ability than previously established multi-generation models.

Quantum process may conduce to brain function. Particularly, it was proposed that the Posner molecules, Ca_g(PO₄)₆, can serve as neural qubits, which can maintain quantum entanglement between phosphorus nuclei for quite a long time. We study the process from entangled Posner molecules to the synchronized activities of brain. A diffusion model of Posner molecules with appropriate boundary conditions and the mean first arriving time of Posner molecules from one area of brain to another are obtained. We establish linkages between the quantum entangled connectivity of brain and the remaining time of entangled Posner molecules before decoherence. Based on these results, we find direct correlations between entangled Posner molecules and the functional connectivity of human brain. Furthermore, we discover that the quantum entangled connectivity depends on age.

We implemented a Markov Chain Monte Carlo technique to estimate the probability-density functions of the cosmic-ray transport and source parameters in a diffusion model. From the measurement of the B/C ratio and radioactive cosmic-ray clocks, we calculate their probability density functions, with a special emphasis on the halo size L of the Galaxy and the local underdense bubble of size r_h. Finally, we check the compatibility of the primary fluxes with the transport parameters derived from the B/C analysis and then derive the source parameters (slope, abundance, and low-energy shape). We conclude that the size of the diffusive halo depends on the presence/absence of the local underdensity damping effect on radioactive nuclei. Models based on fitting B/C are compatible with primary fluxes. The different spectral indices for the propagated primary fluxes up to a few TeV/n can be naturally ascribed to transport effects only, implying universality of elemental source spectra.