Deep learning-based Structure-based drug design aims to generate ligand molecules with desirable properties for protein targets. While existing models have demonstrated competitive performance in generating ligand molecules, they primarily focus on learning the chemical distribution of training datasets, often lacking effective steerability to ensure the desired chemical quality of generated molecules. To address this issue, we propose a multi-reward optimization framework that fine-tunes generative models for attributes, such as binding affinity, validity, and drug-likeness, together. Specifically, we derive direct preference optimization for a Bayesian flow network, used as a backbone for molecule generation, and integrate a reward normalization scheme to adopt multiple objectives. Experimental results show that our method generates more realistic ligands than baseline models while achieving higher binding affinity, expanding the Pareto front empirically observed in previous studies.

Neural networks inspired by differential equations have proliferated for the past several years, of which neural ordinary differential equations (NODEs) and neural controlled differential equations (NCDEs) are two representative examples. In theory, NCDEs exhibit better representation learning capability for time-series data than NODEs. In particular, it is known that NCDEs are suitable for processing irregular time-series data. Whereas NODEs have been successfully extended to adopt attention, methods to integrate attention into NCDEs have not yet been studied. To this end, we present $\underline{\mathrm{A}}$ttentive $\underline{\mathrm{N}}$eural $\underline{\mathrm{C}}$ontrolled $\underline{\mathrm{D}}$ifferential $\underline{\mathrm{E}}$quations (ANCDEs) for time-series classification and forecasting, where dual NCDEs are used: one for generating attention values, and the other for evolving hidden vectors for a downstream machine learning task. We conduct experiments on three real-world time-series datasets and ten baselines. After dropping some values, we also conduct experiments on irregular time-series. Our method consistently shows the best accuracy in all cases by non-trivial margins. Our visualizations also show that the presented attention mechanism works as intended by focusing on crucial information.

This paper reports preliminary work with an RFID based real time location system (RTLS) designed for location and guidance of people in welfare facilities like the sanatorium which has the both of indoor and out door environment. The system consists of a number of fixed RF readers and a number of active RFID tags carried by the target object, or senior people. We compare the performance of several RTLS schemes which use the received signal strength indication (RSSI) values emitted by the moving active RFID tags. Traditional trilateration, fingerprinting, and well-known LANDMARC approaches are evaluated and compared using the SystemC-based computer simulation. Results show a mean estimated error (MEE) performance of 3m with the 130 tags according to the position updated frequency in our simulation environment.

IEEE 802.11e wireless LAN standard supports the call admission control (CAC) for the quality of service (QoS) traffic such as video and voice based on the medium time which means the bandwidth allocation normalized by the arrival time interval of corresponding traffic streams. IEEE 802.11e defines the medium time and its argument, MPDU exchange time which means the duration for the transmission of corresponding traffic stream. The task group (TGe) for 802.11e recommend the definition of MPDU exchange time by using the minimum transmission physical layer rate (PHY rate) without considering adaptive PHY rate to the wireless channel variations. In this paper, we propose new MPDU exchange time considering the adaptive PHY rate instead of fixed rate. The CAC with this proposed MPDU exchange time enhance the user capacity by about 45% maintaining the guarantee of QoS.

The IEEE 802.1 le standard allows an admission control (AC) criteria based on medium time calculation for contention-based wireless access. In this standard, medium time is derived from minimum physical layer (PHY) rate and static surplus bandwidth allocation (SBA) parameters. However, a pessimistic minimum PHY rate and fixed SBA would result in channel underutilization. Therefore, we propose an efficient AC algorithm based on adaptive PHY rate and SBA that enhances the user capacity with quality of service (QoS) provision.

This paper presents a novel high-speed low-complexity pipelined degree-computationless modified Euclidean (pDCME) algorithm architecture for high-speed RS decoders. The pDCME algorithm allows elimination of the degree-computation so as to reduce hardware complexity and obtain high-speed processing. A high-speed RS decoder based on the pDCME algorithm has been designed and implemented with 0.13-μm CMOS standard cell technology in a supply voltage of 1.1 V. The proposed RS decoder operates at a clock frequency of 660 MHz and has a throughput of 5.3 Gb/s. The proposed architecture requires approximately 15% fewer gate counts and a simpler control logic than architectures based on the popular modified Euclidean algorithm.

In this paper, we propose the modified single-path delay feedback (SDF) architecture for FFT implementation, which implements a mixed decimation-in-frequency (DIF)/decimation-in-time (DIT) FFT algorithm. Since final stage is computed as DIT FFT algorithm and other stages including input stage are computed as DIF FFT algorithm, both input and output data occur in normal order and additional clocks for reordering input or output is not required. This architecture is applied to a 64-point FFT and compared to the radix-4 DIF SDF and radix-4 multi-path delay commutator (MDC) architecture in the context of throughput, latency and hardware complexity. As a result, the proposed architecture has a much lower hardware complexity as compared to the radix-4 MDC while maintaining the same throughput and latency, and it achieves a significantly lower latency compared to the original radix-4 SDF architecture with reasonable hardware complexity increment.