wovepaper — the arXiv observatory

The paper presents SonicAGI, a system for real-world target speaker extraction that combines simulated and real meeting data, using a low‑latency SwiftNet-Lookahead model for onlin…

#target speaker extraction#low‑latency online separation#data‑centric training#cross‑attention

cs.CV2026

Difference-Driven Gating: Adaptive Feature Fusion for U-Net Decoder

Kai Li, Xuechao Zou, Jiashen Fu +3

The paper introduces two difference-driven gating mechanisms that adaptively fuse high‑level and low‑level features in U‑Net decoders, improving performance on tasks like medical i…

Publications (130)

Bridging the Sim-to-Real Gap from the Information Bottleneck Perspective

Natural Language Induced Adversarial Images

Enhancing Spectrogram Realism in Singing Voice Synthesis via Explicit Bandwidth Extension Prior to Vocoder

PartImageNet++ Dataset: Enhancing Visual Models with High-Quality Part Annotations

Convolutional Neural Networks with Gated Recurrent Connections

VaccineRAG: Boosting Multimodal Large Language Models' Immunity to Harmful RAG Samples

Estimation of the volume of the left ventricle from MRI images using deep neural networks

Defense against Adversarial Attacks Using High-Level Representation Guided Denoiser

Look Closer to Segment Better: Boundary Patch Refinement for Instance Segmentation

Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection

MiVE: Multiscale Vision-language features for reference-guided video Editing

NP-Match: Towards a New Probabilistic Model for Semi-Supervised Learning

The SonicAGI System for the REAL-TSE Challenge

Difference-Driven Gating: Adaptive Feature Fusion for U-Net Decoder

PMSS: Pretrained Matrices Skeleton Selection for LLM Fine-tuning

DeepInv: A Novel Self-supervised Learning Approach for Fast and Accurate Diffusion Inversion

Amplification trojan network: Attack deep neural networks by amplifying their inherent weakness

Interlinked Convolutional Neural Networks for Face Parsing

Language-Driven Anchors for Zero-Shot Adversarial Robustness

A Brain-Inspired Deep Separation Network for Single Channel Raman Spectra Unmixing

Adversarial Attacks and Defences Competition

Adversarial Texture for Fooling Person Detectors in the Physical World

RSG: A Simple but Effective Module for Learning Imbalanced Datasets

Fooling thermal infrared pedestrian detectors in real world using small bulbs

Evaluate the Malignancy of Pulmonary Nodules Using the 3D Deep Leaky Noisy-or Network

Visual Recognition by Request

AudioTrust: Benchmarking the Multifaceted Trustworthiness of Audio Large Language Models

Selective Kernel Networks

Adjacency constraint for efficient hierarchical reinforcement learning

Speech Separation Using an Asynchronous Fully Recurrent Convolutional Neural Network

NP-Match: When Neural Processes meet Semi-Supervised Learning

SPMamba: State-space model is all you need in speech separation

Audio-Visual Speech Separation in Noisy Environments with a Lightweight Iterative Model

Attack on practical speaker verification system using universal adversarial perturbations

Neural Retrievers are Biased Towards LLM-Generated Content

Chemical knowledge-informed framework for privacy-aware retrosynthesis learning

Pixel Embedding Method for Tubular Neurite Segmentation

MTADiffusion: Mask Text Alignment Diffusion Model for Object Inpainting

IIANet: An Intra- and Inter-Modality Attention Network for Audio-Visual Speech Separation

GlyphMastero: A Glyph Encoder for High-Fidelity Scene Text Editing

Physically Realizable Natural-Looking Clothing Textures Evade Person Detectors via 3D Modeling

NP-SemiSeg: When Neural Processes meet Semi-Supervised Semantic Segmentation

Defending against Patch-Based and Texture-Based Adversarial Attacks with Spectral Decomposition

On the Privacy Effect of Data Enhancement via the Lens of Memorization

Frame Difference-Based Temporal Loss for Video Stylization

Bridging the Gap Between Training and Inference of Bayesian Controllable Language Models

Dynamic Multi-path Neural Network

PBCAT: Patch-based composite adversarial training against physically realizable attacks on object detection

Crafting Physical Adversarial Examples by Combining Differentiable and Physically Based Renders

Drug repositioning for Alzheimer's disease with transfer learning

Interpretable Disentanglement of Neural Networks by Extracting Class-Specific Subnetwork

Accurate and Reliable Predictions with Mutual-Transport Ensemble

Understanding the Disharmony between Dropout and Batch Normalization by Variance Shift

End-to-End Face Parsing via Interlinked Convolutional Neural Networks

Pruning from Scratch

A Semantics-Assisted Video Captioning Model Trained with Scheduled Sampling

A$^2$-LLM: An End-to-end Conversational Audio Avatar Large Language Model

RTFS-Net: Recurrent Time-Frequency Modelling for Efficient Audio-Visual Speech Separation

TDFNet: An Efficient Audio-Visual Speech Separation Model with Top-down Fusion

Time-Frequency-Based Attention Cache Memory Model for Real-Time Speech Separation

HEDNet: A Hierarchical Encoder-Decoder Network for 3D Object Detection in Point Clouds

Turbo Learning Framework for Human-Object Interactions Recognition and Human Pose Estimation

Improving Pedestrian Attribute Recognition With Weakly-Supervised Multi-Scale Attribute-Specific Localization

Extracting Semantic Knowledge from GANs with Unsupervised Learning

An Audio-Visual Speech Separation Model Inspired by Cortico-Thalamo-Cortical Circuits

PhysBrain 1.0 Technical Report

Beyond the Black Box: A Survey on the Theory and Mechanism of Large Language Models

The Winning Solution to the iFLYTEK Challenge 2021 Cultivated Land Extraction from High-Resolution Remote Sensing Image

RefineMask: Towards High-Quality Instance Segmentation with Fine-Grained Features

Scale-Aware Face Detection

Delving Deeper into the Decoder for Video Captioning

FGNet: Leveraging Feature-Guided Attention to Refine SAM2 for 3D EM Neuron Segmentation

SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detection

2D Gaussians Meet Visual Tokenizer

Enhancing In-Context Learning Performance with just SVD-Based Weight Pruning: A Theoretical Perspective

CooHOI: Learning Cooperative Human-Object Interaction with Manipulated Object Dynamics

DanceCrafter: Fine-Grained Text-Driven Controllable Dance Generation via Choreographic Syntax

Rethinking Natural Adversarial Examples for Classification Models

Faster-GCG: Efficient Discrete Optimization Jailbreak Attacks against Aligned Large Language Models