Shushan Qiao

Yuhan Zhang, Zhou Wang, Zhou Shu et al.

In recent years, DeepSeek has achieved strong inference performance but remains hard to deploy on energy-constrained edge devices. This paper presents the DeepSeek Processing Element (DSPE), an edge-oriented architecture that alleviates the model's heavy computational and energy demands. DSPE introduces three techniques: the MerkleTree-based Incremental Pruning Scheme (MIPS) for secure redundant-vector reduction, the Multi-Stage Boothing Lookup Method (MBLM) for bit-flip-aware approximate multiplication, and the Dynamic Adaptive Posit Processing Mechanism (DAPPM), which introduces a new DA-Posit format and its corresponding hardware multiplication architecture. Implemented in TSMC 28nm CMOS, DSPE achieves 109.4 TFLOPS/W energy efficiency compared with state-of-the-art designs and offers a scalable foundation for edge deployment.

Yilei Man, Linhai Xie, Shushan Qiao et al.

Spiking Neural Networks (SNNs) have gained enormous popularity in the field of artificial intelligence due to their low power consumption. However, the majority of SNN methods directly inherit the structure of Artificial Neural Networks (ANN), usually leading to sub-optimal model performance in SNNs. To alleviate this problem, we integrate Neural Architecture Search (NAS) method and propose Multi-Attention Differentiable Architecture Search (MA-DARTS) to directly automate the search for the optimal network structure of SNNs. Initially, we defined a differentiable two-level search space and conducted experiments within micro architecture under a fixed layer. Then, we incorporated a multi-dimensional attention mechanism and implemented the MA-DARTS algorithm in this search space. Comprehensive experiments demonstrate our model achieves state-of-the-art performance on classification compared to other methods under the same parameters with 94.40% accuracy on CIFAR10 dataset and 76.52% accuracy on CIFAR100 dataset. Additionally, we monitored and assessed the number of spikes (NoS) in each cell during the whole experiment. Notably, the number of spikes of the whole model stabilized at approximately 110K in validation and 100k in training on datasets.

Wulve Yang, Hailong Zou, Rui Zhou et al.

General matrix multiplication (GEMM) on spatial accelerators is highly sensitive to mapping choices in both execution efficiency and energy consumption. However, the mapping space exhibits combinatorial explosion, which makes it extremely challenging to obtain optimal mappings within an acceptable time budget. Existing approaches typically face challenges: They often lack global-optimality guarantees and become prohibitively slow as the mapping space grows. To address these limitations, we propose \textsc{GOMA}, a geometric-abstraction-based, globally optimal GEMM mapping framework via analytical modeling, which achieves efficient solving while guaranteeing optimality. \textsc{GOMA} introduces, from first principles, a geometric abstraction for GEMM mapping, yielding an exact analytical energy objective with $O(1)$ evaluation for any given mapping. The objective is highly accurate. \textsc{GOMA} then formulates mapping selection as an integer optimization problem under hardware and mapping constraints, using the analytical energy model as the objective to automate mapping search. \textsc{GOMA} can quickly compute a global-optimal mapping for any (GEMM workload, target hardware) pair, achieving this for the first time in mapping space exploration. Experiments confirm that across representative accelerators and large language model prefill workloads, \textsc{GOMA} improves the energy--delay product (EDP) by $2.24$--$4.24\times$ over SOTA mappers, while accelerating time-to-solution by $3.83$--$73.6\times$.

Ziang Wang, Xiaoqin Wang, Dingyi Wang et al.

Image degradation caused by complex lighting conditions such as low-light and backlit scenarios is commonly encountered in real-world environments, significantly affecting image quality and downstream vision tasks. Most existing methods focus on a single type of illumination degradation and lack the ability to handle diverse lighting conditions in a unified manner. To address this issue, we propose a dual-illumination enhancement framework called DIME-Net. The core of our method is a Mixture-of-Experts illumination estimator module, where a sparse gating mechanism adaptively selects suitable S-curve expert networks based on the illumination characteristics of the input image. By integrating Retinex theory, this module effectively performs enhancement tailored to both low-light and backlit images. To further correct illumination-induced artifacts and color distortions, we design a damage restoration module equipped with Illumination-Aware Cross Attention and Sequential-State Global Attention mechanisms. In addition, we construct a hybrid illumination dataset, MixBL, by integrating existing datasets, allowing our model to achieve robust illumination adaptability through a single training process. Experimental results show that DIME-Net achieves competitive performance on both synthetic and real-world low-light and backlit datasets without any retraining. These results demonstrate its generalization ability and potential for practical multimedia applications under diverse and complex illumination conditions.