Built with and by Teycir Ben Soltane•
How to Use•FAQ•GitHub•arXiv.org•
Share:
ArXivCSExplorer
☆☆Bookmarks🏆RSSHow to UseFAQ
Home/Authors/Xin Zhao

Xin Zhao

8 indexed papers

Recent (6 mo)
7
With code
0
Influential cites
0
Benchmarked
0

Publications per year

8
26

Top categories

Sound×2AI×2Info Retrieval×2Crypto×2Architecture×1NLP×1Audio and Speech Processing×1ML×1

Frequent co-authors

Shixin Zhao1×
Lian Liu1×
Tianhua Han1×
Mengdi Wang1×
Yinhe Han1×
Ying Wang1×

Research Timeline

2026
DPDSyn: Improving Differentially Private Dataset Synthesis for Model Training by Downstream Task Guidance

DPDSyn improves differentially private dataset synthesis by training a differentially private AI model on the original private data, which is then used to generate synthetic datasets that maintain high utility for downstream tasks.

Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures

The paper introduces VIPER, a novel backdoor attack framework that exploits the functional fusion of malicious and benign logic within dynamic prompt architectures, demonstrating a new, high-risk threat.

SPAR: Support-Preserving Action Rectification

SPAR introduces a novel framework that rectifies action policies by performing local fine-tuning in a residual space anchored to a pure behavior cloning policy, achieving state-of-the-art performance in offline policy improvement.

Dual-Stream MLP is All You Need for CTR Prediction

This paper proposes a novel framework, DS-MLP, for click-through rate prediction in online advertising and recommendation systems.

Towards Robust Generative Speech Enhancement Using Vector Quantisation-Based Neural Audio Codec

This paper investigates modelling strategies in continuous and discrete latent spaces in VQ-based neural audio codec speech enhancement.

Multi-Turn Agentic Scientific Literature Search via Workflow Induction

PaperPilot is an interactive literature search agent that constructs an executable DAG of paper-search operators based on user queries and feedback, improving search results and reducing errors.

Re-Sonance: A Dysarthric Asynchronous Real-Time Speech Conversion System Based on a Three-Stage Cascaded ASR-LLM-TTS Architecture

This paper introduces Re-Sonance, a real-time speech-driven AAC system for professional speaking scenarios using LLM-enhanced Whisper ASR, Qwen LLM, and CosyVoice TTS.

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM

This paper proposes PFM, a dual-view memory system for NPU-PIM systems that decouples physical data layout from accessor-visible logical views, reducing bandwidth underutilization and performance loss.

Highlighted terms show continued research focus across papers

Papers

cs.AREmpiricalRecentAug 7, 2026

Rethinking Unified Memory for NPU-PIM Systems: Dual-View Memory for Dynamic Inference of LLM

Shixin Zhao, Lian Liu, Tianhua Han, Mengdi Wang +2 more

This paper proposes PFM, a dual-view memory system for NPU-PIM systems that decouples physical data layout from accessor-visible logical views, reducing bandwidth underutilization and performance loss…

View →
cs.SDcs.AIEmpirical
Recent
Jul 20, 2026

Re-Sonance: A Dysarthric Asynchronous Real-Time Speech Conversion System Based on a Three-Stage Cascaded ASR-LLM-TTS Architecture

Yuxuan Wu, Yifan Xu, Junkun Wang, Jiayong Jiang +2 more

This paper introduces Re-Sonance, a real-time speech-driven AAC system for professional speaking scenarios using LLM-enhanced Whisper ASR, Qwen LLM, and CosyVoice TTS.

View →
cs.CLcs.IREmpiricalRecentJul 1, 2026

Multi-Turn Agentic Scientific Literature Search via Workflow Induction

Jisen Li, Bingxuan Li, Nanyi Jiang, Xuying Ning +9 more

PaperPilot is an interactive literature search agent that constructs an executable DAG of paper-search operators based on user queries and feedback, improving search results and reducing errors.

View →
eess.AScs.SDEmpiricalRecentJun 15, 2026

Towards Robust Generative Speech Enhancement Using Vector Quantisation-Based Neural Audio Codec

Haixin Zhao, Nilesh Madhu

This paper investigates modelling strategies in continuous and discrete latent spaces in VQ-based neural audio codec speech enhancement.

View →
cs.IRRecentJun 3, 2026

Dual-Stream MLP is All You Need for CTR Prediction

Kesha Ou, Zhen Tian, Wayne Xin Zhao, Long Zhang +2 more

This paper proposes a novel framework, DS-MLP, for click-through rate prediction in online advertising and recommendation systems.

View →
cs.LGcs.AIRecentMay 27, 2026

SPAR: Support-Preserving Action Rectification

Jiaxin Zhao, Weihang Pan, Xun Liang, Binbin Lin

SPAR introduces a novel framework that rectifies action policies by performing local fine-tuning in a residual space anchored to a pure behavior cloning policy, achieving state-of-the-art performance…

View →
cs.CRcs.CVMay 19, 2026

Exposing Functional Fusion: A New Class of Strategic Backdoor in Dynamic Prompt Architectures

Zeyao Liu, Zhendong Zhao, Xiaojun Chen, Xin Zhao +2 more

The paper introduces VIPER, a novel backdoor attack framework that exploits the functional fusion of malicious and benign logic within dynamic prompt architectures, demonstrating a new, high-risk thre…

View →
cs.CRApr 17, 2026

DPDSyn: Improving Differentially Private Dataset Synthesis for Model Training by Downstream Task Guidance

Mingxuan Jia, Wen Huang, Weixin Zhao, Xingyi Wang +2 more

DPDSyn improves differentially private dataset synthesis by training a differentially private AI model on the original private data, which is then used to generate synthetic datasets that maintain hig…

View →