Recent advances in Neural Combinatorial Optimization (NCO) methods have significantly improved the capability of neural solvers to handle synthetic routing instances. Nonetheless, existing neural solvers typically struggle to generalize effectively from synthetic, uniformly-distributed training data to real-world VRP scenarios, including widely recognized benchmark instances from TSPLib and CVRPLib. To bridge this generalization gap, we present Evolutionary Realistic Instance Synthesis (EvoReal), which leverages an evolutionary module guided by large language models (LLMs) to generate synthetic instances characterized by diverse and realistic structural patterns. Specifically, the evolutionary module produces synthetic instances whose structural attributes statistically mimics those observed in authentic real-world instances. Subsequently, pre-trained NCO models are progressively refined, firstly aligning them with these structurally enriched synthetic distributions and then further adapting them through direct fine-tuning on actual benchmark instances. Extensive experimental evaluations demonstrate that EvoReal markedly improves the generalization capabilities of state-of-the-art neural solvers, yielding a notable reduced performance gap compared to the optimal solutions on the TSPLib (1.05%) and CVRPLib (2.71%) benchmarks across a broad spectrum of problem scales.

In the graph model for conflict resolution (GMCR), option prioritization is an important approach to obtain preferences of decision makers (DMs) over a special conflict. The preference, based on ordered preference statements, is influenced by the authenticity and the order of preference statements. Owing to ambiguous evidence and incomplete information in complex real-world conflicts, it is of great difficulty to determine crisp preference over states for DMs. Some uncertain preferences have been introduced to cope with two situations that uncertain about preference statements authenticity or prioritization. The belief structure can be used to capture vagueness or unknown in subjective judgments. In this article, we propose a belief option prioritization technique by considering two situations of preference statements to elicit preferences comprehensively and efficiently. First, a belief structure is applied to describe the uncertainty of preference statement authenticity. The belief degree represents the accuracy of each preference statement and the real preference attitude of a DM. Second, a belief distribution, associated with an ordered sequence of preference statements, is used to capture the uncertainty on the prioritization of preference statements. The belief preference over feasible states can be obtained by the scoring scheme of option prioritization based on the ordered preference statements. Last, we propose an overall belief option prioritization technique by combining two uncertain situations. The application of Gisborne Lake water export conflict is utilized to illustrate the use of belief option prioritization.

Sorting solutions play a key role in using evolutionary algorithms (EAs) to solve many-objective optimization problems (MaOPs). Generally, different solution-sorting methods possess different advantages in dealing with distinct MaOPs. Focusing on this characteristic, this article proposes a general voting-mechanism-based ensemble framework (VMEF), where different solution-sorting methods can be integrated and work cooperatively to select promising solutions in a more robust manner. In addition, a strategy is designed to calculate the contribution of each solution-sorting method and then the total votes are adaptively allocated to different solution-sorting methods according to their contribution. Solution-sorting methods that make more contribution to the optimization process are rewarded with more votes and the solution-sorting methods with poor contribution will be punished in a period of time, which offers a good feedback to the optimization process. Finally, to test the performance of VMEF, extensive experiments are conducted in which VMEF is compared with five state-of-the-art peer many-objective EAs, including NSGA-III, SPEA/R, hpaEA, BiGE, and grid-based evolutionary algorithm. Experimental results demonstrate that the overall performance of VMEF is significantly better than that of these comparative algorithms.

Convolution operation accounts for the major work-load in convolutional neural networks (CNN). However, standard instruction set for RISC-V processor cannot efficiently perform the matrix convolution between kernel and input matrices. In this paper, we construct a custom instruction under the RISC-V ISA that can perform the F(2×2,3×3) convolution within one single execution. Particularly, optimized by the Winograd algorithm, the operation only needs 16 multiplications instead of 36 multiplications as needed by standard ISA. Benefit from this cycles, as compared to 140 cycles using standard instructions. Thenew instruction, F(2×2,3×3) can be calculated within 19 clock power consumed during convolution operation is also reduced significantly.

Probabilistic shaping is widely used in optical communication to increase the data rate. Because the number of symbols is an integer, when modulating the input signals into the predefined symbol sequence, the modulated probabilistic distribution is a discrete M-type distribution. In this paper, a fast method of M-type approximation based on information divergence is proposed. Compared with traditional algorithm, the proposed fast M-type approximation can greatly reduce the computation complexity while achieving the same results. This paper also proves mathematically that the information divergence increment has boundaries for M-type approximation near the ideal distribution. Based on the bounds of information divergence increment, the M-type approximation distribution boundary related to ideal distribution can be obtained.

Although Artificial Neural Networks (ANNs) are inspired by biological neural systems, most of ANNs today are implemented with digital circuitry and use binary values in computation. In recent years, analog-based neuromorphic system has gained lots of attention as it provides a natural interface for brain-machine interaction. In this paper, we present analog designs of a complete neuron system, where the Multiply-Accumulate (MAC) and Rectified Linear Unit (ReLU) functions are all implemented in analog circuits. The design uses SMIC 55nm standard LP CMOS process node and operates at low supply voltage (1.2 V). The simulation results in SPECTRE demonstrate that the MAC's linear error is no more than 0.5% and total harmonic distortion (THD) is less than 1.6% when the inputs vary from peak (-10 µA) to peak (10 µA) at 10 MHz, the -3dB bandwidth is 288 MHz, the maximum power consumption is 540 µW and the static power consumption is 493 µW under 100MHz input signal frequency. More specifically, our design is resilient to the fluctuation of power supply, which helps to achieve high precision of computation.

Spiking Neural Networks (SNN) is one special implementation of Artificial Neural Networks (ANN), where the input signals are encoded in the temporal relationship between consecutive spikes (spike trains) instead of real-numbered values. Nevertheless, SNN is believed to be a closer representation of the biological neural system, because it imitates the current spikes that are transmitted between neurons in real biological systems. Practical and simplified SNN models include the Leaky Integrate-and-Fire (LIF) function of the neurons and the Spiking Timing Dependent Plasticity (STDP) function of the synapses. While most ANN architectures can be implemented with digital logic gates, SNN is more suitable to be implemented in analog circuits. In this paper, we propose revised analog circuit implementations of SNN neurons with LIF and STDP functions. Compared with previous works by other researchers, our proposed analog designs use fewer components and can be cascaded to form a complete neural system. The circuits are designed and simulated with SMIC 55nm CMOS LP process. The simulated results demonstrate that the analog neural system can work under a very small current (less than 10 µA) and voltage supply (1.0 Volts), and consumes less power consumption than digital implementations.

Nowadays, providing quality of service (QoS) guarantees for some applications such as signal data processing has become a critical issue. In this paper, we propose a novel self-adaptive QoS-aware scheduling algorithm called SAQA that sufficiently considers the adaptability for real-time tasks with QoS demands on heterogeneous clusters. When the system is in heavy load, the SAQA algorithm can degrade the QoS levels of new tasks or tasks waiting in local queues of nodes to guarantee high schedulability. The minimum QoS level is acceptable for each task. In contrast, when the system is in light load, SAQA can use slack time to adequately improve the QoS of new tasks. We compare SAQA with SAEDF algorithm by simulations. The experimental results indicate that SAQA has admirable adaptability while providing timing and QoS guarantees.

Electro-magnetic Detection Satellite (EDS) is an important branch of Earth Observation Satellites (EOSs). It has been widely applied in industry and military areas. The detecting duration of EDS is different from imagery satellites, for it is an imprecise parameter because of the uncertain electromagnetic environment within space surrounding targets. This factor makes the task scheduling for EDS becoming a complex combinatorial optimization problem. With consideration of this special property, we used fuzzy set and possibility theory to model imprecise parameters, other physical constraints, like on-board energy and transition time between different working patterns and sensor's rebooting, were also taken into account in the model. We presented an improved genetic algorithm to solve this problem by introducing new method of elitist and parents selection. The model and the algorithm have been tested by five experiments derived from STK's satellite database.