1. 9.5 Feature Matching Data Synthesis for Non-IID Federated Learning
   • Authors: Zijian Li, Yuchang Sun, Jiawei Shao, Yuyi Mao, Jessie Hui Wang, Jun Zhang
   • Reason: The authors present a new method for tackling the non-IID issue in Federated Learning by synthesizing data that matches class-relevant features. The novel FL framework, integrated with the proposed method, demonstrates advantage in accuracy, privacy preservation, and computational cost.
2. 9.3 Intrinsic Motivation via Surprise Memory
   • Authors: Hung Le, Kien Do, Dung Nguyen, Svetha Venkatesh
   • Reason: The authors innovatively propose an original model for intrinsic rewards in reinforcement learning that promises more efficient exploring behaviors and boosts performance in sparse reward environments.
3. 9.0 Tram-FL: Routing-based Model Training for Decentralized Federated Learning
   • Authors: Kota Maejima, Takayuki Nishio, Asato Yamazaki, Yuko Hara-Azumi
   • Reason: The paper presents a novel decentralized federated learning method, Tram-FL, that refines a global model by sequentially transferring it amongst nodes to improve performance and reduce communication costs.
4. 8.7 When and How Does Known Class Help Discover Unknown Ones? Provable Understanding Through Spectral Analysis
   • Authors: Yiyou Sun, Zhenmei Shi, Yingyu Liang, Yixuan Li
   • Reason: The paper introduces a new analytical framework to understand when and how known classes can aid in the discovery of novel classes with a theoretically guaranteed approach.
5. 8.6 Collaborative Learning From Distributed Data With Differentially Private Synthetic Twin Data
   • Authors: Lukas Prediger, Joonas Jälkö, Antti Honkela, Samuel Kaski
   • Reason: The authors make headway in privacy-preserving collaborative learning methods by proposing a synthesizing data with differential privacy framework enabling learning from sensitive data without violating privacy constraints.