改进前向-前向学习，提升模型稳定性和泛化能力
Adaptive Multi-Scale Goodness Aggregation for Forward-Forward Learning
- 多尺度好坏度聚合，融合局部到全局特征
- 在MNIST和Fashion-MNIST上分别提升1.45%和1.50%
- 适合追求高效且生物合理训练机制的研究者
我们提出自适应多尺度好坏度聚合（AMSGA），一种改进的前向-前向（FF）算法，旨在提升局部学习神经网络的稳定性、鲁棒性与泛化能力。AMSGA通过在局部、中间和全局表征间进行多尺度好坏度聚合；引入基于课程引导的困难负样本挖掘；采用层依赖的自适应阈值；以及使用余弦退火预热学习率策略，增强优化稳定性。这些改进在保持原有生物合理性与内存效率的同时，显著强化了FF范式。在MNIST和Fashion-MNIST上的实验表明，相比基线FF算法，性能持续提升，分别达+1.45%和+1.50%，且计算开销几乎无增加。结果表明，当好坏度估计与训练动态被精心设计时，局部学习方法可具备更强竞争力。
原文摘要 · Abstract (English)
We propose Adaptive Multi-Scale Goodness Aggregation (AMSGA), a novel extension of the Forward-Forward (FF) algorithm designed to improve stability, robustness, and generalization in local-learning neural networks. AMSGA addresses several limitations of the original FF framework by introducing multi-scale goodness aggregation across local, intermediate, and global representations; adaptive curriculum-guided hard negative mining; layer-dependent adaptive thresholds; and a warm-up cosine annealing learning-rate schedule for improved optimization stability. Together, these modifications strengthen the FF paradigm while preserving its biologically plausible and memory-efficient properties. Experiments on MNIST and Fashion-MNIST demonstrate consistent performance improvements over the baseline FF algorithm, achieving up to +1.45% improvement on MNIST and +1.50% improvement on Fashion-MNIST without significant computational overhead. Our results suggest that local learning methods can become substantially more competitive when goodness estimation and training dynamics are carefully designed.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。