arXiv subjects

Feng Zhang

Publications and source records attributed to Feng Zhang.

At least 109 records · Page 6Linked to original sources

arXiv2025

Asymmetric Diffusion Recommendation Model

Recently, motivated by the outstanding achievements of diffusion models, the diffusion process has been employed to strengthen representation learning in recommendation systems. Most diffusion-based recommendation models typically utilize standard Gaussian noise in symmetric forward and reverse processes in continuous data space. Nevertheless, the samples derived from recommendation systems inhabit a discrete data space, which is fundamentally different from the continuous one. Moreover, Gaussian noise has the potential to corrupt personalized information within latent representations. In this work, we propose a novel and effective method, named Asymmetric Diffusion Recommendation Model (AsymDiffRec), which learns forward and reverse processes in an asymmetric manner. We define a generalized forward process that simulates the missing features in real-world recommendation samples. The reverse process is then performed in an asymmetric latent feature space. To preserve personalized information within the latent representation, a task-oriented optimization strategy is introduced. In the serving stage, the raw sample with missing features is regarded as a noisy input to generate a denoising and robust representation for the final prediction. By equipping base models with AsymDiffRec, we conduct online A/B tests, achieving improvements of +0.131% and +0.166% in terms of users' active days and app usage duration respectively. Additionally, the extended offline experiments also demonstrate improvements. AsymDiffRec has been implemented in the Douyin Music App.

cs.IR↗

arXiv2025

RankMixer: Scaling Up Ranking Models in Industrial Recommenders

Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMixer retains the transformer's high parallelism while replacing quadratic self-attention with multi-head token mixing module for higher efficiency. Besides, RankMixer maintains both the modeling for distinct feature subspaces and cross-feature-space interactions with Per-token FFNs. We further extend it to one billion parameters with a Sparse-MoE variant for higher ROI. A dynamic routing strategy is adapted to address the inadequacy and imbalance of experts training. Experiments show RankMixer's superior scaling abilities on a trillion-scale production dataset. By replacing previously diverse handcrafted low-MFU modules with RankMixer, we boost the model MFU from 4.5\% to 45\%, and scale our ranking model parameters by 100x while maintaining roughly the same inference latency. We verify RankMixer's universality with online A/B tests across two core application scenarios (Recommendation and Advertisement). Finally, we launch 1B Dense-Parameters RankMixer for full traffic serving without increasing the serving cost, which improves user active days by 0.3\% and total in-app usage duration by 1.08\%.

cs.IR↗

arXiv2025

Robust Low-light Scene Restoration via Illumination Transition

Synthesizing normal-light novel views from low-light multiview images is an important yet challenging task, given the low visibility and high ISO noise present in the input images. Existing low-light enhancement methods often struggle to effectively preprocess such low-light inputs, as they fail to consider correlations among multiple views. Although other state-of-the-art methods have introduced illumination-related components offering alternative solutions to the problem, they often result in drawbacks such as color distortions and artifacts, and they provide limited denoising effectiveness. In this paper, we propose a novel Robust Low-light Scene Restoration framework (RoSe), which enables effective synthesis of novel views in normal lighting conditions from low-light multiview image inputs, by formulating the task as an illuminance transition estimation problem in 3D space, conceptualizing it as a specialized rendering task. This multiview-consistent illuminance transition field establishes a robust connection between low-light and normal-light conditions. By further exploiting the inherent low-rank property of illumination to constrain the transition representation, we achieve more effective denoising without complex 2D techniques or explicit noise modeling. To implement RoSe, we design a concise dual-branch architecture and introduce a low-rank denoising module. Experiments demonstrate that RoSe significantly outperforms state-of-the-art models in both rendering quality and multiview consistency on standard benchmarks. The codes and data are available at https://pegasus2004.github.io/RoSe.

cs.CV↗

arXiv2025

Balancing Semantic Relevance and Engagement in Related Video Recommendations

Related video recommendations commonly use collaborative filtering (CF) driven by co-engagement signals, often resulting in recommendations lacking semantic coherence and exhibiting strong popularity bias. This paper introduces a novel multi-objective retrieval framework, enhancing standard two-tower models to explicitly balance semantic relevance and user engagement. Our approach uniquely combines: (a) multi-task learning (MTL) to jointly optimize co-engagement and semantic relevance, explicitly prioritizing topical coherence; (b) fusion of multimodal content features (textual and visual embeddings) for richer semantic understanding; and (c) off-policy correction (OPC) via inverse propensity weighting to effectively mitigate popularity bias. Evaluation on industrial-scale data and a two-week live A/B test reveals our framework's efficacy. We observed significant improvements in semantic relevance (from 51% to 63% topic match rate), a reduction in popular item distribution (-13.8% popular video recommendations), and a +0.04% improvement in our topline user engagement metric. Our method successfully achieves better semantic coherence, balanced engagement, and practical scalability for real-world deployment.

cs.IR↗

arXiv2025

Lighting the Night with Generative Artificial Intelligence

The visible light reflectance data from geostationary satellites is crucial for meteorological observations and plays an important role in weather monitoring and forecasting. However, due to the lack of visible light at night, it is impossible to conduct continuous all-day weather observations using visible light reflectance data. This study pioneers the use of generative diffusion models to address this limitation. Based on the multi-band thermal infrared brightness temperature data from the Advanced Geostationary Radiation Imager (AGRI) onboard the Fengyun-4B (FY4B) geostationary satellite, we developed a high-precision visible light reflectance generative model, called Reflectance Diffusion (RefDiff), which enables 0.47~μ\mathrm{m}, 0.65~μ\mathrm{m}, and 0.825~μ\mathrm{m} bands visible light reflectance generation at night. Compared to the classical models, RefDiff not only significantly improves accuracy through ensemble averaging but also provides uncertainty estimation. Specifically, the SSIM index of RefDiff can reach 0.90, with particularly significant improvements in areas with complex cloud structures and thick clouds. The model's nighttime generation capability was validated using VIIRS nighttime product, demonstrating comparable performance to its daytime counterpart. In summary, this research has made substantial progress in the ability to generate visible light reflectance at night, with the potential to expand the application of nighttime visible light data.

cs.CV↗

arXiv2025

Targeting Melanoma-Specific Tyrosinase: Cyclic Peptide Disrupts Actin Dynamics for Precision Apoptosis Induction

Melanoma is an aggressive and highly metastatic cancer that exhibits stubborn resistance to conventional therapies, highlighting the need for novel treatments. Existing therapeutic strategies often suffer from systemic toxicity, poor efficacy and fast-gained drug resistance. In this study, we designed a cyclic peptide system (c-RGDKYQ) that takes the advantage of the overexpression of tyrosinase in melanoma cells to trigger enzyme-mediated oxidation and self-assembly. The assembled peptide nanostructures can selectively disrupt the actin cytoskeleton, impairing cancer cellular functions, e.g., motility, adhesion, and proliferation, ultimately leading to apoptosis. This approach does not rely on external drug payloads or complex delivery mechanisms. c-RGDKYQ exhibits high selectivity for melanoma cells, strongly suppressing tumor growth in a murine model with minimal systemic toxicity. Our findings illuminate that, through targeting tyrosinase, c-RGDKYQ may be an enzyme-responsive alternative to conventional treatments for melanoma.

q-bio.BM↗

arXiv2025

Towards Solving More Challenging IMO Problems via Decoupled Reasoning and Proving

Automated Theorem Proving (ATP) in formal languages is a foundational challenge for AI. While Large Language Models (LLMs) have driven remarkable progress, a significant gap remains between their powerful informal reasoning capabilities and their weak formal proving performance. Recent studies show that the informal accuracy exceeds 80% while formal success remains below 8% on benchmarks like PutnamBench. We argue this gap persists because current state-of-the-art provers, by tightly coupling reasoning and proving, are trained with paradigms that inadvertently punish deep reasoning in favor of shallow, tactic-based strategies. To bridge this fundamental gap, we propose a novel framework that decouples high-level reasoning from low-level proof generation. Our approach utilizes two distinct, specialized models: a powerful, general-purpose Reasoner to generate diverse, strategic subgoal lemmas, and an efficient Prover to rigorously verify them. This modular design liberates the model's full reasoning potential and bypasses the pitfalls of end-to-end training. We evaluate our method on a challenging set of post-2000 IMO problems, a problem set on which no prior open-source prover has reported success. Our decoupled framework successfully solves 5 of these problems, demonstrating a significant step towards automated reasoning on exceptionally difficult mathematical challenges. To foster future research, we release our full dataset of generated and verified lemmas for a wide range of IMO problems, available at https://tencent-imo.github.io/ .

cs.LO↗

arXiv2025

Hunyuan-TurboS: Advancing Large Language Models through Mamba-Transformer Synergy and Adaptive Chain-of-Thought

As Large Language Models (LLMs) rapidly advance, we introduce Hunyuan-TurboS, a novel large hybrid Transformer-Mamba Mixture of Experts (MoE) model. It synergistically combines Mamba's long-sequence processing efficiency with Transformer's superior contextual understanding. Hunyuan-TurboS features an adaptive long-short chain-of-thought (CoT) mechanism, dynamically switching between rapid responses for simple queries and deep "thinking" modes for complex problems, optimizing computational resources. Architecturally, this 56B activated (560B total) parameter model employs 128 layers (Mamba2, Attention, FFN) with an innovative AMF/MF block pattern. Faster Mamba2 ensures linear complexity, Grouped-Query Attention minimizes KV cache, and FFNs use an MoE structure. Pre-trained on 16T high-quality tokens, it supports a 256K context length and is the first industry-deployed large-scale Mamba model. Our comprehensive post-training strategy enhances capabilities via Supervised Fine-Tuning (3M instructions), a novel Adaptive Long-short CoT Fusion method, Multi-round Deliberation Learning for iterative improvement, and a two-stage Large-scale Reinforcement Learning process targeting STEM and general instruction-following. Evaluations show strong performance: overall top 7 rank on LMSYS Chatbot Arena with a score of 1356, outperforming leading models like Gemini-2.0-Flash-001 (1352) and o4-mini-2025-04-16 (1345). TurboS also achieves an average of 77.9% across 23 automated benchmarks. Hunyuan-TurboS balances high performance and efficiency, offering substantial capabilities at lower inference costs than many reasoning models, establishing a new paradigm for efficient large-scale pre-trained models.

cs.CL↗

arXiv2025

$C^3$-Bench: The Things Real Disturbing LLM based Agent in Multi-Tasking

Agents based on large language models leverage tools to modify environments, revolutionizing how AI interacts with the physical world. Unlike traditional NLP tasks that rely solely on historical dialogue for responses, these agents must consider more complex factors, such as inter-tool relationships, environmental feedback and previous decisions, when making choices. Current research typically evaluates agents via multi-turn dialogues. However, it overlooks the influence of these critical factors on agent behavior. To bridge this gap, we present an open-source and high-quality benchmark $C^3$-Bench. This benchmark integrates attack concepts and applies univariate analysis to pinpoint key elements affecting agent robustness. In concrete, we design three challenges: navigate complex tool relationships, handle critical hidden information and manage dynamic decision paths. Complementing these challenges, we introduce fine-grained metrics, innovative data collection algorithms and reproducible evaluation methods. Extensive experiments are conducted on 49 mainstream agents, encompassing general fast-thinking, slow-thinking and domain-specific models. We observe that agents have significant shortcomings in handling tool dependencies, long context information dependencies and frequent policy-type switching. In essence, $C^3$-Bench aims to expose model vulnerabilities through these challenges and drive research into the interpretability of agent performance. The benchmark is publicly available at https://github.com/TencentHunyuan/C3-Benchmark.

cs.AI↗

arXiv2025

exa-AMD: A Scalable Workflow for Accelerating AI-Assisted Materials Discovery and Design

exa-AMD is a Python-based application designed to accelerate the discovery and design of functional materials by integrating AI/ML tools, materials databases, and quantum mechanical calculations into scalable, high-performance workflows. The execution model of exa-AMD relies on Parsl, a task-parallel programming library that enables a flexible execution of tasks on any computing resource from laptops to supercomputers. By using Parsl, exa-AMD is able to decouple the workflow logic from execution configuration, thereby empowering researchers to scale their workflows without having to reimplement them for each system.

cs.DC↗

arXiv2025

Endpoint estimates for the fractal circular maximal function and related local smoothing

Sharp $L^p$--$L^q$ estimates for the spherical maximal function over dilation sets of fractal dimensions, including the endpoint estimates, were recently proved by Anderson--Hughes--Roos--Seeger. More intricate $L^p$--$L^q$ estimates for the fractal circular maximal function were later established in the sharp range by Roos--Seeger, but the endpoint estimates have been left open, particularly when the fractal dimension of the dilation set lies in $[1/2, 1)$. In this work, we prove these missing endpoint estimates for the circular maximal function. We also study the closely related $L^p$--$L^q$ local smoothing estimates for the wave operator over fractal dilation sets, which were recently investigated by Beltran--Roos--Rutar--Seeger and Wheeler. Making use of a bilinear approach, we also extend the range of $p,q$, for which the optimal estimate holds.

math.CA↗

arXiv2025

Pyramid Mixer: Multi-dimensional Multi-period Interest Modeling for Sequential Recommendation

Sequential recommendation, a critical task in recommendation systems, predicts the next user action based on the understanding of the user's historical behaviors. Conventional studies mainly focus on cross-behavior modeling with self-attention based methods while neglecting comprehensive user interest modeling for more dimensions. In this study, we propose a novel sequential recommendation model, Pyramid Mixer, which leverages the MLP-Mixer architecture to achieve efficient and complete modeling of user interests. Our method learns comprehensive user interests via cross-behavior and cross-feature user sequence modeling. The mixer layers are stacked in a pyramid way for cross-period user temporal interest learning. Through extensive offline and online experiments, we demonstrate the effectiveness and efficiency of our method, and we obtain a +0.106% improvement in user stay duration and a +0.0113% increase in user active days in the online A/B test. The Pyramid Mixer has been successfully deployed on the industrial platform, demonstrating its scalability and impact in real-world applications.

cs.IR↗

arXiv2025

HARMONY: A Scalable Distributed Vector Database for High-Throughput Approximate Nearest Neighbor Search

Approximate Nearest Neighbor Search (ANNS) is essential for various data-intensive applications, including recommendation systems, image retrieval, and machine learning. Scaling ANNS to handle billions of high-dimensional vectors on a single machine presents significant challenges in memory capacity and processing efficiency. To address these challenges, distributed vector databases leverage multiple nodes for the parallel storage and processing of vectors. However, existing solutions often suffer from load imbalance and high communication overhead, primarily due to traditional partition strategies that fail to effectively distribute the workload. In this paper, we introduce Harmony, a distributed ANNS system that employs a novel multi-granularity partition strategy, combining dimension-based and vector-based partition. This strategy ensures a balanced distribution of computational load across all nodes while effectively minimizing communication costs. Furthermore, Harmony incorporates an early-stop pruning mechanism that leverages the monotonicity of distance computations in dimension-based partition, resulting in significant reductions in both computational and communication overhead. We conducted extensive experiments on diverse real-world datasets, demonstrating that Harmony outperforms leading distributed vector databases, achieving 4.63 times throughput on average in four nodes and 58% performance improvement over traditional distribution for skewed workloads.

cs.DB↗

arXiv2025

Learning Unpaired Image Dehazing with Physics-based Rehazy Generation

Overfitting to synthetic training pairs remains a critical challenge in image dehazing, leading to poor generalization capability to real-world scenarios. To address this issue, existing approaches utilize unpaired realistic data for training, employing CycleGAN or contrastive learning frameworks. Despite their progress, these methods often suffer from training instability, resulting in limited dehazing performance. In this paper, we propose a novel training strategy for unpaired image dehazing, termed Rehazy, to improve both dehazing performance and training stability. This strategy explores the consistency of the underlying clean images across hazy images and utilizes hazy-rehazy pairs for effective learning of real haze characteristics. To favorably construct hazy-rehazy pairs, we develop a physics-based rehazy generation pipeline, which is theoretically validated to reliably produce high-quality rehazy images. Additionally, leveraging the rehazy strategy, we introduce a dual-branch framework for dehazing network training, where a clean branch provides a basic dehazing capability in a synthetic manner, and a hazy branch enhances the generalization ability with hazy-rehazy pairs. Moreover, we design a new dehazing network within these branches to improve the efficiency, which progressively restores clean scenes from coarse to fine. Extensive experiments on four benchmarks demonstrate the superior performance of our approach, exceeding the previous state-of-the-art methods by 3.58 dB on the SOTS-Indoor dataset and by 1.85 dB on the SOTS-Outdoor dataset in PSNR. Our code will be publicly available.

cs.CV↗

arXiv2025

Optimizing Recall or Relevance? A Multi-Task Multi-Head Approach for Item-to-Item Retrieval in Recommendation

The task of item-to-item (I2I) retrieval is to identify a set of relevant and highly engaging items based on a given trigger item. It is a crucial component in modern recommendation systems, where users' previously engaged items serve as trigger items to retrieve relevant content for future engagement. However, existing I2I retrieval models in industry are primarily built on co-engagement data and optimized using the recall measure, which overly emphasizes co-engagement patterns while failing to capture semantic relevance. This often leads to overfitting short-term co-engagement trends at the expense of long-term benefits such as discovering novel interests and promoting content diversity. To address this challenge, we propose MTMH, a Multi-Task and Multi-Head I2I retrieval model that achieves both high recall and semantic relevance. Our model consists of two key components: 1) a multi-task learning loss for formally optimizing the trade-off between recall and semantic relevance, and 2) a multi-head I2I retrieval architecture for retrieving both highly co-engaged and semantically relevant items. We evaluate MTMH using proprietary data from a commercial platform serving billions of users and demonstrate that it can improve recall by up to 14.4% and semantic relevance by up to 56.6% compared with prior state-of-the-art models. We also conduct live experiments to verify that MTMH can enhance both short-term consumption metrics and long-term user-experience-related metrics. Our work provides a principled approach for jointly optimizing I2I recall and semantic relevance, which has significant implications for improving the overall performance of recommendation systems.

cs.IR↗

arXiv2025

Quantum-Classical Embedding via Ghost Gutzwiller Approximation for Enhanced Simulations of Correlated Electron Systems

Simulating correlated materials on present-day quantum hardware remains challenging due to limited quantum resources. Quantum embedding methods offer a promising route by reducing computational complexity through the mapping of bulk systems onto effective impurity models, allowing more feasible simulations on pre- and early-fault-tolerant quantum devices. This work develops a quantum-classical embedding framework based on the ghost Gutzwiller approximation to enable quantum-enhanced simulations of ground-state properties and spectral functions of correlated electron systems. Circuit complexity is analyzed using an adaptive variational quantum algorithm on a statevector simulator, applied to the infinite-dimensional Hubbard model with increasing ghost mode numbers from 3 to 5, resulting in circuit depths growing from 16 to 104. Noise effects are examined using a realistic error model, revealing significant impact on the spectral weight of the Hubbard bands. To mitigate these effects, the Iceberg quantum error detection code is employed, achieving up to 40% error reduction in simulations. Finally, the accuracy of the density matrix estimation is benchmarked on IBM and Quantinuum quantum hardware, featuring distinct qubit-connectivity and employing multiple levels of error mitigation techniques.

quant-ph↗

arXiv2025

Revealing Fano Resonance in Dirac Materials ZrTe5 through Raman Scattering

We explore the Fano resonance in ZrTe5, using Raman scattering measurements. We identified two closely spaced B2g phonon modes, B2g I and B2g II, around 9 meV and 10 meV, respectively. Interestingly, only B2g I exhibited the Fano resonance, an outcome of quantum interference between discrete phonon modes and continuous electronic excitations. This is consistent with the much stronger electron-phonon coupling of B2g I mode demonstrated by first-principles calculations. Additionally, temperature-dependent measurements highlight an enhanced Fano asymmetry at elevated temperatures, contributed by the thermal renormalization of the band structure and electron-phonon coupling. This study offers insights into the complex interrelation of electron-phonon coupling, thermal effects, and Fano resonances in ZrTe5.

cond-mat.mtrl-sci↗

arXiv2025

Adaptive Deep Reasoning: Triggering Deep Thinking When Needed

Large language models (LLMs) have shown impressive capabilities in handling complex tasks through long-chain reasoning. However, the extensive reasoning steps involved can significantly increase computational costs, posing challenges for real-world deployment. Recent efforts have focused on optimizing reasoning efficiency by shortening the Chain-of-Thought (CoT) reasoning processes through various approaches, such as length-aware prompt engineering, supervised fine-tuning on CoT data with variable lengths, and reinforcement learning with length penalties. Although these methods effectively reduce reasoning length, they still necessitate an initial reasoning phase. More recent approaches have attempted to integrate long-chain and short-chain reasoning abilities into a single model, yet they still rely on manual control to toggle between short and long CoT. In this work, we propose a novel approach that autonomously switches between short and long reasoning chains based on problem complexity. Our method begins with supervised fine-tuning of the base model to equip both long-chain and short-chain reasoning abilities. We then employ reinforcement learning to further balance short and long CoT generation while maintaining accuracy through two key strategies: first, integrating reinforcement learning with a long-short adaptive group-wise reward strategy to assess prompt complexity and provide corresponding rewards; second, implementing a logit-based reasoning mode switching loss to optimize the model's initial token choice, thereby guiding the selection of the reasoning type. Evaluations on mathematical datasets demonstrate that our model can dynamically switch between long-chain and short-chain reasoning modes without substantially sacrificing performance. This advancement enhances the practicality of reasoning in large language models for real-world applications.

cs.CL↗