Qiwen Wang

Huanxin Yang, Qiwen Wang

Handwritten mathematical expression recognition (HMER) suffers from complex formula structures and character layouts in sequence prediction. In this paper, we incorporate frequency domain analysis into HMER and propose a method that marries frequency domain with HMER (MFH), leveraging the discrete cosine transform (DCT). We emphasize the structural analysis assistance of frequency information for recognizing mathematical formulas. When implemented on various baseline models, our network exhibits a consistent performance enhancement, demonstrating the efficacy of frequency domain information. Experiments show that our MFH-CoMER achieves noteworthy accuracyrates of 61.66%/62.07%/63.72% on the CROHME 2014/2016/2019 test sets. The source code is available at https://github.com/Hryxyhe/MFH.

Yong Deng, Guoqing Wang, Zhenzhe Ying et al.

Large language models (LLMs) exhibit remarkable problem-solving abilities, but struggle with complex tasks due to static internal knowledge. Retrieval-Augmented Generation (RAG) enhances access to external information, yet remains limited in multi-hop reasoning and strategic search due to rigid workflows. Recent advancements in agentic deep research empower LLMs to autonomously reason, search, and synthesize information. However, current approaches relying on outcome-based reinforcement learning (RL) face critical issues such as conflicting gradients and reward sparsity, limiting performance gains and training efficiency. To address these, we first propose Atomic Thought, a novel LLM thinking paradigm that decomposes reasoning into fine-grained functional units. These units are supervised by Reasoning Reward Models (RRMs), which provide Atomic Thought Rewards (ATR) for fine-grained guidance. Building on this, we propose Atom-Searcher, a novel RL framework for agentic deep research that integrates Atomic Thought and ATR. Atom-Searcher uses a curriculum-inspired reward schedule, prioritizing process-level ATR early and transitioning to outcome rewards, accelerating convergence on effective reasoning paths. Experiments on seven benchmarks show consistent improvements over the state-of-the-art. Key advantages include: (1) Atom-Searcher scales computation at test-time. (2) Atomic Thought provides supervision anchors for RRMs, bridging deep research tasks and RRMs. (3) Atom-Searcher exhibits more interpretable, human-like reasoning patterns.

Yachuan Huang, Xianrui Luo, Qiwen Wang et al.

Bokeh rendering simulates the shallow depth-of-field effect in photography, enhancing visual aesthetics and guiding viewer attention to regions of interest. Although recent approaches perform well, rendering controllable bokeh without additional depth inputs remains a significant challenge. Existing classical and neural controllable methods rely on accurate depth maps, while generative approaches often struggle with limited controllability and efficiency. In this paper, we propose BokehFlow, a depth-free framework for controllable bokeh rendering based on flow matching. BokehFlow directly synthesizes photorealistic bokeh effects from all-in-focus images, eliminating the need for depth inputs. It employs a cross-attention mechanism to enable semantic control over both focus regions and blur intensity via text prompts. To support training and evaluation, we collect and synthesize four datasets. Extensive experiments demonstrate that BokehFlow achieves visually compelling bokeh effects and offers precise control, outperforming existing depth-dependent and generative methods in both rendering quality and efficiency.

Yi Luo, Qiwen Wang, Junqi Yang et al.

Generalized Category Discovery (GCD) aims to classify both known and novel categories using partially labeled data that contains only known classes. Despite achieving strong performance on existing benchmarks, current textual GCD methods lack sufficient validation in realistic settings. We introduce Event-Centric GCD (EC-GCD), characterized by long, complex narratives and highly imbalanced class distributions, posing two main challenges: (1) divergent clustering versus classification groupings caused by subjective criteria, and (2) Unfair alignment for minority classes. To tackle these, we propose PaMA, a framework leveraging LLMs to extract and refine event patterns for improved cluster-class alignment. Additionally, a ranking-filtering-mining pipeline ensures balanced representation of prototypes across imbalanced categories. Evaluations on two EC-GCD benchmarks, including a newly constructed Scam Report dataset, demonstrate that PaMA outperforms prior methods with up to 12.58% H-score gains, while maintaining strong generalization on base GCD datasets.

Yuting Wu, Qiwen Wang, Ziyu Wang et al.

The need for deep neural network (DNN) models with higher performance and better functionality leads to the proliferation of very large models. Model training, however, requires intensive computation time and energy. Memristor-based compute-in-memory (CIM) modules can perform vector-matrix multiplication (VMM) in situ and in parallel, and have shown great promises in DNN inference applications. However, CIM-based model training faces challenges due to non-linear weight updates, device variations, and low-precision in analog computing circuits. In this work, we experimentally implement a mixed-precision training scheme to mitigate these effects using a bulk-switching memristor CIM module. Lowprecision CIM modules are used to accelerate the expensive VMM operations, with high precision weight updates accumulated in digital units. Memristor devices are only changed when the accumulated weight update value exceeds a pre-defined threshold. The proposed scheme is implemented with a system-on-chip (SoC) of fully integrated analog CIM modules and digital sub-systems, showing fast convergence of LeNet training to 97.73%. The efficacy of training larger models is evaluated using realistic hardware parameters and shows that that analog CIM modules can enable efficient mix-precision DNN training with accuracy comparable to full-precision software trained models. Additionally, models trained on chip are inherently robust to hardware variations, allowing direct mapping to CIM inference chips without additional re-training.

Yingbai Hu, Jian Li, Yongquan Chen et al.

The outbreak of novel coronavirus pneumonia (COVID-19) has caused mortality and morbidity worldwide. Oropharyngeal-swab (OP-swab) sampling is widely used for the diagnosis of COVID-19 in the world. To avoid the clinical staff from being affected by the virus, we developed a 9-degree-of-freedom (DOF) rigid-flexible coupling (RFC) robot to assist the COVID-19 OP-swab sampling. This robot is composed of a visual system, UR5 robot arm, micro-pneumatic actuator and force-sensing system. The robot is expected to reduce risk and free up the clinical staff from the long-term repetitive sampling work. Compared with a rigid sampling robot, the developed force-sensing RFC robot can facilitate OP-swab sampling procedures in a safer and softer way. In addition, a varying-parameter zeroing neural network-based optimization method is also proposed for motion planning of the 9-DOF redundant manipulator. The developed robot system is validated by OP-swab sampling on both oral cavity phantoms and volunteers.

Qiwen Wang, Hua Sun, Mikael Skoglund

We consider the problem of private information retrieval (PIR) with colluding servers and eavesdroppers (abbreviated as ETPIR). The ETPIR problem is comprised of $K$ messages, $N$ servers where each server stores all $K$ messages, a user who wants to retrieve one of the $K$ messages without revealing the desired message index to any set of $T$ colluding servers, and an eavesdropper who can listen to the queries and answers of any $E$ servers but is prevented from learning any information about the messages. The information theoretic capacity of ETPIR is defined to be the maximum number of desired message symbols retrieved privately per information symbol downloaded. We show that the capacity of ETPIR is $C = \left( 1- \frac{E}{N} \right) \left(1 + \frac{T-E}{N-E} + \cdots + \left( \frac{T-E}{N-E} \right)^{K-1} \right)^{-1}$ when $E < T$, and $C = \left( 1 - \frac{E}{N} \right)$ when $E \geq T$. To achieve the capacity, the servers need to share a common random variable (independent of the messages), and its size must be at least $\frac{E}{N} \cdot \frac{1}{C}$ symbols per message symbol. Otherwise, with less amount of shared common randomness, ETPIR is not feasible and the capacity reduces to zero. An interesting observation is that the ETPIR capacity expression takes different forms in two regimes. When $E < T$, the capacity equals the inverse of a sum of a geometric series with $K$ terms and decreases with $K$; this form is typical for capacity expressions of PIR. When $E \geq T$, the capacity does not depend on $K$, a typical form for capacity expressions of SPIR (symmetric PIR, which further requires data-privacy, {\it i.e.,} the user learns no information about other undesired messages); the capacity does not depend on $T$ either. In addition, the ETPIR capacity result includes multiple previous PIR and SPIR capacity results as special cases.

Qiwen Wang, Mikael Skoglund

A user wants to retrieve a file from a database without revealing the identity of the file retrieved at the database, which is known as the problem of private information retrieval (PIR). If it is further required that the user obtains no information about the database other than the desired file, the concept of symmetric private information retrieval (SPIR) is introduced to guarantee privacy for both parties. In this paper, the problem of SPIR is studied for a database stored among $N$ nodes in a distributed way, by using an $(N,M)$-MDS storage code. The information-theoretic capacity of SPIR, defined as the maximum number of symbols of the desired file retrieved per downloaded symbol, for the coded database is derived. It is shown that the SPIR capacity for coded database is $1-\frac{M}{N}$, when the amount of the shared common randomness of distributed nodes (unavailable at the user) is at least $\frac{M}{N-M}$ times the file size. Otherwise, the SPIR capacity for the coded database equals zero.