1. 9.7 NUPES : Non-Uniform Post-Training Quantization via Power Exponent Search
   • Authors: Edouard Yvinec, Arnaud Dapogny, Kevin Bailly
   • The authors propose a non-uniform quantization technique to tackle limitations of standard deep learning model quantization. This approach allows optimization of quantization operator itself during training and ensures compatibility with integer-only low-bit inference, achieving superior compression rates.
2. 9.5 Homophily-enhanced Structure Learning for Graph Clustering
   • Authors: Ming Gu, Gaoming Yang, Sheng Zhou, Ning Ma, Jiawei Chen, Qiaoyu Tan, Meihan Liu, Jiajun Bu
   • The paper presents a novel approach for graph clustering based on homophily-enhanced structure learning. The proposed method outperforms state-of-the-art baselines in experiments on a variety of benchmark datasets.
3. 9.3 IIHT: Medical Report Generation with Image-to-Indicator Hierarchical Transformer
   • Authors: Keqiang Fan, Xiaohao Cai, Mahesan Niranjan
   • The authors proposed a system that uses a hierarchical transformer to generate medical reports, potentially easing the workload on doctors. This novel approach demonstrated promising results on a diverse set of metrics and provides a hook for integration in real-world scenarios and human-AI collaboration.
4. 9.2 Flexible Isosurface Extraction for Gradient-Based Mesh Optimization
   • Authors: Tianchang Shen, Jacob Munkberg, Jon Hasselgren, Kangxue Yin, Zian Wang, Wenzheng Chen, Zan Gojcic, Sanja Fidler, Nicholas Sharp, Jun Gao
   • The paper introduces FlexiCubes, a new isosurface representation for optimizing unknown mesh in terms of geometric, visual, or physical objectives. This new paradigm shows significant improvements in mesh quality and geometric fidelity and extends the current state of the art in this domain.
5. 9.1 Updating Clinical Risk Stratification Models Using Rank-Based Compatibility: Approaches for Evaluating and Optimizing Clinician-Model Team Performance
   • Authors: Erkin Ötleş, Brian T. Denton, Jenna Wiens
   • The authors propose a new rank-based compatibility measure and a loss function for updating clinical risk stratification models. This novel approach to model updating shows promise for improving user-model team performance in healthcare settings.