Amil Aligayev

Deep learning in medical imaging faces a persistent tension between preserving localized diagnostic information and maintaining computational efficiency. We introduce saccadic lightweight efficient attention (SLEA), a framework purpose-built for 2D medical image classification and inspired by selective visual processing in the human visual system. Rather than dividing an image into fixed patches or processing it at once, SLEA independently samples multiple localized crops from uniformly distributed spatial coordinates, with sampled locations re-drawn at every training epoch to provide spatially diverse exposure without requiring a learned fixation policy. Each glimpse is encoded through a shared ImageNet-pretrained MobileNetV2 backbone into a compact feature token; the resulting unordered token set is integrated through multi-head self-attention followed by mean pooling, capturing content-level relationships among spatially distributed regions within a minimal parameter budget. This stochastic sampling is intended to reduce dependence on fixed image locations and recurring spatial artefacts, although this regularization effect is an empirical hypothesis rather than a formally established mechanism. Experiments on ChestX-ray14, ISIC2019, and APTOS2019 show strong performance under the evaluated protocols while maintaining a compact model size of approximately 2.45 million parameters. However, the ChestX-ray14 result is substantially higher than the commonly reported benchmark range and should be interpreted cautiously; independent replication under closely matched partitions, preprocessing, optimization, and inference protocols remains necessary. Results from the COVID-19 Radiography Database are reported separately, since publicly assembled COVID-19 datasets may contain source-, acquisition-, and processing-related differences that create artificial class separability; this result is therefore excluded from the principal performance claims and is not interpreted as evidence of clinical diagnostic capability or general superiority. A multi-granularity interpretability scheme combining glimpse-level Grad-CAM, top-glimpse, and multi-glimpse visualizations provides qualitative indications of the regions associated with the model’s predictions, but these are not interpreted as exact lesion-localization maps or definitive evidence of clinically faithful reasoning.

This study investigates the electronic and photocatalytic properties of greenly fabricated rutile-phase titanium dioxide (bTiO₂) modified with iron vanadate (Fe/bTiO₂/VO₄) and vanadium-substituted goethite (Fe/bTiO₂/VOOH) by detailed experimental assay and density functional theory (DFT) calculations. Our analysis of the density of states (DOS), band structure, and work function reveals that both dopant systems significantly modify the electronic structure of pure rutile bTiO₂. The dye methyl orange (MO) was used as the model pollutant. During photodegradation tests, parameters such as the reaction time, solid-to-liquid ratio, initial concentrations of the photocatalyst and dye, as well as distance of the lamp from the reactor and pH were varied. Degradation kinetics follows the equation of the pseudo-first order law for both photocatalysts (k_VO4 = 0.058 min⁻¹ and k_VOOH = 0.065 min⁻¹), while degradation efficiencies of 92% and 99% were observed after 120 min at pH 3, respectively. Specifically, the DOS analysis highlights the contribution of Fe 3d and V 3d orbitals, which create new electronic states within the bandgap, facilitating charge transfer. These insights provide a strong foundation for the rational design of novel, highly efficient Fe/bTiO₂-based photocatalysts for the degradation of organic pollutants in water.

This computational study delves into the intricate interplay of alloying elements on the generation, recombination, and evolution of irradiation-induced defects. Molecular dynamics simulations were conducted for collision cascades at room temperature, spanning a range of primary knock-on atom energies from 1 to 10 keV. The investigation encompasses a series of model crystals, progressing from pure Ni to binary concentrated solid solution alloys (CSAs) such as NiFe₂₀, NiFe, NiCr₂₀, and NiFeCr₂₀ CSA. We observe that materials rich in Cr actively facilitate dislocation emissions and induce the nucleation of stacking fault tetrahedra in the proximity of nanovoids, due to Shockley partial interactions. This result is validated by molecular static simulations, which calculate the surface, vacancy, and defect formation energies. Among the various shapes considered, the spherical void proves to be the most stable, followed by the truncated octahedron and octahedron shapes. On the other hand, the tetrahedron cubic shape is identified as the most unstable, and stacking fault tetrahedra exhibit the highest formation energy. Notably, among the materials studied, NiCr₂₀ and NiFeCr₂₀ CSAs stood out as the sole alloys capable of manifesting this mechanism, mainly observed at high impact energies.

Herein, the physisorption mechanisms of , , , and molecules on alumina and their effect on electronic properties are investigated. Quantum–classical molecular dynamics simulations and the self‐consistent‐charge density‐functional tight‐binding approach are used to dynamically model these mechanisms. Herein, the binding pathways of O, H, and C atoms in the various molecules to Al and O atoms at the top atomic layers of the ‐alumina surface are revealed by the results. Several adsorption sites and molecular orientations relative to Al‐terminated and Ox‐terminated alumina surfaces are examined and it is found that the most stable physisorbed state on the Al‐terminated surface is located above the Al atom, while the Ox‐terminated state is found above the oxygen, resulting in enhanced optical adsorbance. The dissociation of into after interaction with the surface results in hydrogen production, but with low adsorbate rates, while molecules primarily bond to the Al atoms, leading to the highest adsorbance rate among the other molecules. Herein, important insights are provided by the findings into the physisorption mechanisms of molecules on alumina and their impact on electronic properties.