Hanoi University of Science and Technology

This study aims to develop an AI prediction system for watersheds, utilizing data from Sri Lanka and Vietnam in collaboration with overseas researchers. Initially, a foundational AI prediction system will be constructed, and basic data will be gathered based on river data from Japan. Subsequently, topographic data of the watershed area, managed by the joint research partner institution, will be incorporated into the AI prediction system through drone-based aerial measurement and analysis. Moreover, visits to the joint research institution will be conducted to gather information on water damage and electricity supply. This information will then be integrated into the AI prediction system. Utilizing the collected data and additional observation data, efforts will be made to enhance the accuracy of the AI prediction algorithm, identify potential issues, and overcome them. The ultimate goal is to establish new AI prediction technology by comparing the results of this study with both domestic and international research outcomes. This paper will present data from a drone scan of the upstream watershed and discuss the future use of AI based on the results of three dimensional analysis and image processing

Objective. The International Commission on Radiological Protection (ICRP) decided to develop pregnant-female reference computational phantoms, including the maternal and fetal phantoms, through its 2007 general recommendations. Acknowledging the advantages of the mesh geometry, the ICRP decided to develop the pregnant-female mesh-type reference computational phantoms (MRCPs) for 8, 10, 15, 20, 25, 30, 35, and 38 week fetal ages directly in the mesh format. As part of this process, the present study developed the mesh-type fetal phantoms. Approach. The reference blood-inclusive organ masses, elemental compositions, and densities were established based on various scientific literatures. Then, the phantoms were developed in accordance with the established reference dataset while reflecting the anatomical features of the developing fetus, such as fetal-age-specific anthropometric parameters, bone ossification, and contents formation time. Main results. The phantoms were developed in the tetrahedral-mesh format and can be implemented in the general-purpose Monte Carlo codes (i.e. Geant4, PHITS, MCNP6, and EGSnrc) without the necessity of the voxelization process. To explore the dosimetric impact of the developed phantoms, photon specific absorbed fractions (SAFs) were computed for energies between 10⁻²–10¹ MeV for the fetal liver and spleen as source regions and self-irradiation and cross-irradiation to the fetal brain, lungs, and urinary bladder wall as target regions. The SAFs showed the fetal-age-dependent dose trends (i.e. SAF decreases with increasing fetal age) due to organ masses increases via fetal growth. Significance. The mesh-type fetal phantoms, as part of the ICRP pregnant-female MRCPs, will be used to calculate reference dose coefficients for fetal members of the public for both the current and future ICRP general recommendations.

The hole expansion ratio (HER) observed in a standardized hole expansion test (HET) is commonly used to determine the edge fracture of steel sheets. A large variation of the measured HER restricts the practical application of the method. This study presents a systematic investigation on uncertainties in the HER of DP800 sheet material, including the hole-edge quality, pre-strain due to the hole-punching process, the friction coefficient, and the determination of fracture. An artificial neural network was trained to develop a surrogate model using a database gained from a thousand finite element simulations of the HET. Monte-Carlo simulations were performed using the trained surrogate model to characterize the distribution of the HER. Sensitivity analysis via Sobol’s indices is calculated to determine the influence of the input variables on the output. It is found that the pre-strain and pre-damage generated during the hole punching process in the shear-affected zone dominate the variation of the HER. Discussions on reducing the output’s variation are detailed. In general, these findings provide valuable insights for the determination of HER as well as the edge crack behavior of the investigated DP800 steel sheet.

There are several types of propolis in Brazil produced by Apis mellifera Linnaeus, 1758, Apidae, with the green propolis from the Caatinga biome standing out for its high flavonoid content. In this study, we describe the isolation of flavonoids from Brazilian green propolis of Caatinga Mimosa tenuiflora (Willd.) Poir., Fabaceae, and the development of a reliable RP-HPLC quantitative method. This method uses a Shim-pack VP-ODS column (250 × 4.6 mm i.d., 5 μm) with nonlinear gradient elution and UV detection at 280 nm. Additionally, a sample preparation method for extracting flavonoids using 96% ethanol and caffeic acid as the internal standard was employed. The developed method demonstrated excellent detection response, with limits of detection and quantification ranging from 0.65–2.08 µg/ml and 1.97–6.31 µg/ml, respectively. The maximum relative standard deviation was 4.61%. Thirteen flavonoids were quantified, including santin, ermanin, sakuranetin, quercetin-3-methyl ether, viscosine, eriodictyol-5-O-methyl ether, isokaempferide, kaempferide, penduletin, quercetin-3,6,7-trimethyl ether, cirsimaritin, 3,3'-O-dimethylquercetin, and luteolin. The developed method met all the parameters set by international guidelines for analytical method development. It is reliable for the quality control of M. tenuiflora green propolis and its related products.

Polyamorphism, the existence of multiple amorphous states in a single material, has been observed in the glass-forming system GeO2. This study investigates the intermediate range order, two-state model and polyamorphism in GeO2 system using molecular dynamics (MD) simulation and MD data mining analytics. Analysis of the Ge-Ge distance distribution revealed the presence of distinct high-density (HD) and low-density (LD) regions in the GeO4 tetrahedral network. The number of clusters was optimized using the silhouette score. Spatial mapping of the HD and LD regions indicated their non-uniform distribution. The distributions of atomic pair distances, O–Ge–O bond angles, and Ge–O–Ge bond angles differ between the HD and LD regions, supporting the two-state model and polyamorphism in GeO2. These insights into the intermediate-range order enhance understanding of the structural origins of GeO2 polyamorphism.

Hydrophobicity is crucial for the interaction between amphipathic antimicrobial peptides and microbial pathogens. However, it is difficult to fully understand the impact of this factor because the biological functions are also influenced by other structural properties, including peptide length, net charge, hydrophilicity, secondary structure, and hydrophobic moment. This study compares three natural antimicrobial peptides—mastoparan C, mastoparan-AF, and mastoparan L—where hydrophobicity varies but other structural features remain nearly identical. Mastoparan C, the most hydrophobic peptide, displays the highest helical content and hemolytic activity, whereas mastoparan-AF, with slightly lower hydrophobicity, demonstrates superior selectivity. In contrast, mastoparan L, the least hydrophobic peptide, exhibits the weakest antimicrobial potency and lowest hemolytic activity, despite showing the least self-assembly. Overall, this study suggests that optimal hydrophobicity, rather than the highest value, enhances antimicrobial efficacy while minimizing hemolytic activity. Graphic abstract Optimal hydrophobicity enhances antimicrobial potency while minimizing hemolytic activity

This research investigates the low cycle fatigue (LCF) and tensile properties of X20CrMoV12.1 steel which belongs to the group 12Cr ferritic steels at 538–566 °C. The tensile strength measured at elevated temperature exhibits a reduction of approximately 30% compared to values obtained at ambient conditions. Conversely, ductility properties show a significant increase under elevated temperatures. Fatigue strength at elevated temperatures is generally lower than at ambient conditions for similar strain amplitudes, with a pronounced reduction observed at low strain amplitudes. Fractographic analysis indicates that the observed reduction in fatigue strength at low strains is due to oxidation-induced microcrack initiations following prolonged exposure to high temperatures. The Basquin-Coffin-Masson models and Ramberg–Osgood models were utilized to characterize the LCF properties of the materials for each temperature condition. The transition fatigue life of sample performed at ambient air is approximately half that of those tested at elevated temperatures. The findings demonstrate that the influence of chromium content on improving the LCF properties of X20CrMoV12.1 material is insignificant compared to materials with lower chromium content under the elevated temperature range of 538–566 °C.

Recommendation systems play a crucial role in providing web-based suggestion utilities by leveraging user behavior, preferences, and interests. In the context of privacy concerns and the proliferation of handheld devices, federated recommender systems have emerged as a promising solution. These systems allow each client to train a local model and exchange only the model updates with a central server, thus preserving data privacy. However, certain use cases necessitate the deduction of contributions from specific clients, a process known as “unlearning”. Existing machine unlearning methods are designed for centralized settings and do not cater to the collaborative nature of recommendation systems, thereby overlooking their unique characteristics. This paper proposes CFRU, a novel federated recommendation unlearning model that enables efficient and certified removal of target clients from the global model. Instead of retraining the model, our approach rolls back and eliminates the historical updates associated with the target client. To efficiently store the learning process's historical updates, we propose sampling strategies that reduce the number of historical updates, retaining only the most significant ones. Furthermore, we analyze the potential bias introduced by the removal of target clients’ updates at each training round and establish an estimation using the Lipschitz condition. Leveraging this estimation, we propose an efficient iterative scheme to accumulate the bias across all rounds, compensating for the removed updates from the global model and recovering its utility without requiring post-training steps. Extensive experiments conducted on two real-world datasets, incorporating two poison attack scenarios, have shown that our unlearning technique can achieve a model quality that is 99.3% equivalent to retraining the model from scratch while performing up to 1000 times faster.

This letter presents a novel design of single-layer differential-fed dual-polarized antenna with low-profile, wideband, and high-isolation characteristics. The antenna is composed of a main cross-shape patch connected to four symbiotic patches by stepped stubs. The patches are symmetrically loaded with mushroom structures to generate extra resonant modes, characterized as TM $_{30}$ , TM $_{12}$ , TM $_{50}$ , and slot modes, which are combined with the TM $_{10}$ mode of the main patch for the wide bandwidth. Thanks to the differential feed and structural symmetry, a very high isolation is achieved among the ports. The final design with low profile of $0.08\lambda _{L}$ ( $\lambda _{L}$ is the free-space wavelength at the lowest operational frequency) has been fabricated and measured. The measurements result in a bandwidth of 51.9% (6.0 $-$ 10.2 GHz) for 10-dB return loss and an isolation of $\ge$ 42 dB. The dual-polarized broadside radiation is validated by far-field measurements showing an average gain of 8.6 dBi and cross-polarization level of $\le -20$ dB within the bandwidth.

This article introduces novel and deep learning approaches for the security analysis of a hybrid satellite-terrestrial cooperative network. More specifically, a satellite transmits information to a ground user through multiple relays in the presence of an eavesdropper. To prevent potential eavesdropping, multiple friendly jammers are employed to disrupt the reception process of the eavesdropper by artificial noise. Within this setting, we then derive the closed-form expressions of the outage probability (OP) and secrecy outage probability (SOP) of the considered system in the presence of imperfect channel state information. Important to mention is the fact that in complex systems (e.g., with multiple jammers, multiple relays, and considering the independent but nonidentically distributed Rician nature of satellite links), analytical approaches may not be effective due to their complex mathematical derivations. As such, we develop a highly effective yet low-complexity deep learning approach to estimate the OP and SOP of the system. Through extensive Monte Carlo simulations, we evaluate the OP and SOP of the system in various settings and demonstrate the effectiveness of the proposed solutions. Interestingly, the proposed deep learning method can achieve comparable performance to that of the analytical approach.

This paper proposes a simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-empowered ambient backscatter short-packet paradigm with partial non-orthogonal multiple access (p–NOMA). In this paradigm, a multi-antenna base station communicates with users using finite blocklength schemes to achieve low latency transmission while flexibly exploiting the spectrum utilization via p–NOMA. Considering Nakagami– m fading channels, discrete phase-shift alignment, and imperfect successive interference cancellation, we provide a generalized information-theoretic framework that characterizes passive, active, and hybrid STAR-RIS types, to measure the block-error rate (BLER) and goodput. To gain useful insights into system designs, an upper-bound BLER at high transmit power has been derived. Numerical results demonstrate the BLER superiority of p–NOMA over its orthogonal multiple access (OMA) and NOMA counterparts, as well as the respective twofold and fourfold enhancements in terms of goodput.

Frequency-coded chipless radio frequency identification (RFID) tag design frequently suffers from the mutual coupling effect among resonant components, which causes the frequency shifting phenomenon. It is very difficult to thoroughly resolve this problem because of design requirements, such as high capacity and limited space on a single tag. In this letter, a novel design method based on particle swarm optimization (PSO) in combination with empirical Taguchi method (TM) is proposed. With this methodology, the optimal design parameters are automatically searched and fitted to comply with the resonance requirements at the given encoding frequencies. The proposed method was demonstrated with the design process of an I-shaped slot tag structure.

This research exploits the applications of reconfigurable intelligent surface (RIS)-assisted multiple input multiple output (MIMO) systems, specifically addressing the enhancement of communication reliability with modulated signals. Specifically, we first derive the analytical downlink symbol error rate (SER) of each user as a multivariate function of both the phase-shift and beamforming vectors. The analytical SER enables us to obtain insights into the synergistic dynamics between the RIS and MIMO communication. We then introduce a novel average SER minimization problem subject to the practical constraints of the transmitted power budget and phase shift coefficients, which is NP-hard. By incorporating the differential evolution (DE) algorithm as a pivotal tool for optimizing the intricate active and passive beamforming variables in RIS-assisted communication systems, the non-convexity of the considered SER optimization problem can be effectively handled. Furthermore, an efficient local search is incorporated into the DE algorithm to overcome the local optimum, and hence offer low SER and high communication reliability. Monte Carlo simulations validate the analytical results and the proposed optimization framework, indicating that the joint active and passive beamforming design is superior to the other benchmarks.

The monolayer $\hbox {C}_4 \hbox {N}_3$BN, short for s-triazine g-$\hbox {C}_4\hbox {N}_3$ with B and pyrrolic N tailored at vacant sites, is investigated under in-plane strain, with a focus on its magnetism. Our density functional study shows that the half-metallic ground state and ferrimagnetic order persist up to considerable strain from −7% to 10%, and temperatures as high as 900 K. At the critical uniaxial strain of approximately −9.30% and 10.90%, or symmetric biaxial strain of −7.62% and 17.38%, this ferrimagnetic half-metal turns into ferromagnetic semiconductors, all with the magnetization 1 \upmu B per unit cell. We present a simple explanation for that metamagnetism in the system by means of spin charge transfer and by counting its atomic valencies in the unit cell. Our finding adds to and enriches physicochemical understanding of the magnetism in carbon nitride-based half-metals.

We determine infinitesimal $\textrm{CR}$ automorphisms and stability groups of certain real hypersurfaces in $\mathbb {C}^2$ in the case when the hypersurface is nonminimal at the reference point.

The present study investigated structural phase transitions in heat-treated mechanical alloyed Al65Cu20Fe15ÿxCrx (x = 2.5, 5, and 10). The Al, Cu, Fe, and Cr powder mixtures were mechanically alloyed by using a planetary ball mill for 30, 60, and 120 minutes. After 120 minutes of milling, Al65Cu20Fe12.5Cr2.5 and Al65Cu20Fe5Cr10 alloys showed b-AlFe(Cu) solid solution phase, and Al65Cu20Fe10Cr5 alloy showed s-AlFe(CuCr) solid solution phase, but no quasicrystalline phase formed during the process. Elemental powders were milled for 30 minutes and heat-treated at 600 °C and 700 °C for 4 hours, resulting in solid-state reactions and forming ω, β, and i phases. Among these, the Al65Cu20Fe12.5Cr2.5 alloy powders had the highest i-phase content after annealing at 700 °C. SEM micrographs of samples annealed at 700 °C for 4 hours exhibited the formation of an icosahedron-shaped quasicrystalline phase with a particle size of about 1 to 2 μm. The powders showed soft ferromagnetic behavior at room temperature.