Structured3D dataset automatically extracts multi-structure geometric relations from professional interior designs, improving indoor scene understanding by 15% on benchmarks.

Jia Zheng, Junfei Zhang, Jing Li et al.

Output/feature distillation and encoder freezing raise incremental VOC segmentation mIoU to 71.6% without storing old images.

Umberto Michieli, Pietro Zanuttigh

GENESIS couples object-centric decomposition with autoregressive scene generation; GQN FID 80.5, ShapeStacks ARI 0.73.

Martin Engelcke, Adam R. Kosiorek, Oiwi Parker Jones et al.

Proposed ReDR network uses dynamic reasoning and reinforcement learning to improve conversational question generation, outperforming baselines on CoQA.

Boyuan Pan, Hao Li, Ziyu Yao et al.

This study leverages pre-trained BERT, GPT-2, RoBERTa checkpoints in a Transformer seq2seq model, achieving state-of-the-art results in translation and summarization.

Sascha Rothe, Shashi Narayan, Aliaksei Severyn

TREQS model uses deep learning to generate SQL queries, enhancing question answering efficiency on electronic medical records.

Ping Wang, Tian Shi, Chandan K. Reddy

Proposes a variational multimodal representation learning framework integrating RGB-D, force, and proprioception, boosting contact-rich task sample efficiency by 20%.

Michelle A. Lee, Yuke Zhu, Peter Zachares et al.

Proposes a cost- and fidelity-layered adaptive sampling strategy for multi-fidelity Gaussian processes, significantly reducing predictive uncertainty.

Sayan Ghosh, Jesper Kristensen, Yiming Zhang et al.

Optuna, a define-by-run hyperparameter optimization framework, enhances flexibility and efficiency with dynamic search spaces and asynchronous pruning.

Takuya Akiba, Shotaro Sano, Toshihiko Yanase et al.

Proposes InterFaceGAN, using linear hyperplanes in latent space for controllable face attribute editing.

Yujun Shen, Jinjin Gu, Xiaoou Tang et al.

Proposed multi-scale local planar guidance layers significantly enhance monocular depth estimation accuracy, outperforming existing methods on NYU Depth V2 and KITTI datasets.

Jin Han Lee, Myung-Kyu Han, Dong Wook Ko et al.

MT-FwFM model enhances conversion prediction via multi-task learning, improving AUC by 0.74% and 0.84%.

Junwei Pan, Yizhi Mao, Alfonso Lobos Ruiz et al.

Wannier90 v3.0 introduces symmetry-adapted, selective localization, and many-body property modules, boosting efficiency and applicability.

Giovanni Pizzi, Valerio Vitale, Ryotaro Arita et al.

Proposes Green AI by emphasizing efficiency metrics like FPO, reducing training costs and carbon footprint.

Roy Schwartz, Jesse Dodge, Noah A. Smith et al.

ELI5 introduces a large-scale long-form QA dataset with multi-source web data, trained with multi-task Seq2Seq models outperforming traditional baselines.

Angela Fan, Yacine Jernite, Ethan Perez et al.

Proposes human-annotated frame shuffling to identify action classes requiring temporal cues, creating a 'Temporal Dataset' for benchmarking temporal modeling.

Laura Sevilla-Lara, Shengxin Zha, Zhicheng Yan et al.

Weak Adversarial Networks (WAN) solve high-dimensional PDEs via weak form and operator norm minimization, offering fast, mesh-free solutions.

Yaohua Zang, Gang Bao, Xiaojing Ye et al.

DeepProbLog integrates neural networks with probabilistic logic programming, enabling symbolic and subsymbolic inference, program induction, and end-to-end training, advancing neuro-symbolic AI.

Robin Manhaeve, Sebastijan Dumančić, Angelika Kimmig et al.

Using semiclassical analysis, the paper derives sharp asymptotics for small eigenvalues of the Witten Laplacian with boundary critical points, highlighting boundary geometry's influence.

Dorian Le Peutrec, Boris Nectoux

Adversarial filtration created ImageNet-A/O; DenseNet-121 reached only ~2% on A and OOD detection was near chance.

Dan Hendrycks, Kevin Zhao, Steven Basart et al.