近期论文20
- xLSTM-Mixer: Multivariate Time Series Forecasting by Mixing via Scalar Memories第 4 位,共 4 位作者NeurIPS 2025 · 27 次引用
- Exploring Neural Granger Causality with xLSTMs: Unveiling Temporal Dependencies in Complex Data第 3 位,共 4 位作者NeurIPS 2025 · 5 次引用
- Finding NeMo: Localizing Neurons Responsible For Memorization in Diffusion Models第 3 位,共 5 位作者NeurIPS 2024 · 46 次引用
- Be Careful What You Smooth For: Label Smoothing Can Be a Privacy Shield but Also a Catalyst for Model Inversion Attacks第 3 位,共 3 位作者ICLR 2024 · 26 次引用
- Learning Large DAGs is Harder than you Think: Many Losses are Minimal for the Wrong DAG第 5 位,共 5 位作者ICLR 2024 · 9 次引用
- Graph Neural Networks Need Cluster-Normalize-Activate Modules第 5 位,共 5 位作者NeurIPS 2024 · 8 次引用
- Rickrolling the Artist: Injecting Backdoors into Text Encoders for Text-to-Image Synthesis第 3 位,共 3 位作者ICCV 2023 · 65 次引用
- Interpretable and Explainable Logical Policies via Neurally Guided Symbolic Abstraction第 5 位,共 5 位作者NeurIPS 2023 · 64 次引用
- ATMAN: Understanding Transformer Predictions Through Memory Efficient Attention Manipulation第 6 位,共 7 位作者NeurIPS 2023 · 45 次引用
- Characteristic Circuits第 3 位,共 3 位作者NeurIPS 2023 · 8 次引用
- ILLUME: Rationalizing Vision-Language Models through Human Interactions第 4 位,共 4 位作者ICML 2023 · 8 次引用
- Do Not Marginalize Mechanisms, Rather Consolidate!第 5 位,共 5 位作者NeurIPS 2023 · 3 次引用
- Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Models第 4 位,共 6 位作者CVPR 2023
- CLEVA-Compass: A Continual Learning Evaluation Assessment Compass to Promote Research Transparency and Comparability第 4 位,共 4 位作者ICLR 2022 · 40 次引用
- Interactive Disentanglement: Learning Concepts by Interacting with their Prototype Representations第 4 位,共 4 位作者CVPR 2022 · 14 次引用
- Right for Better Reasons: Training Differentiable Models by Constraining their Influence Functions第 5 位,共 5 位作者AAAI 2021 · 45 次引用
- Interventional Sum-Product Networks: Causal Inference with Tractable Probabilistic Models第 5 位,共 5 位作者NeurIPS 2021 · 42 次引用
- Right for the Right Concept: Revising Neuro-Symbolic Concepts by Interacting With Their Explanations第 3 位,共 3 位作者CVPR 2021
- Padé Activation Units: End-to-end Learning of Flexible Activation Functions in Deep Networks第 3 位,共 3 位作者ICLR 2020 · 116 次引用
- Structured Object-Aware Physics Prediction for Video Modeling and Planning第 5 位,共 5 位作者ICLR 2020 · 77 次引用
