arXiv cs.AI· Alireza Rahmansetayesh, Amirhossein Yousefiramandi, Maxwell Spiridonov, Sumit Raut, Kent Ziti, Pierre Baldi·· 3 天前AI 评分18
评估 Forward-Forward 学习的新指标：Boundary Proximity Index 与错误决策区域内的样本拯救
Rescuing Points Trapped Inside Incorrect Decision Regions: A target for Evaluating and Improving Forward-Forward Learning
AI 导读
研究者提出 Boundary Proximity Index（BPI）分析框架，用于评估 Forward-Forward 学习及其变体在分类准确率之外的表现。实验发现，未训练的卷积神经网络线性探针准确率已远超随机水平，而靠近决策边界的误分类样本比深陷错误决策区域的样本更容易被"拯救"。该框架在七种 Forward-Forward 变体上验证，揭示了不同方法在拯救深度嵌入错误区域样本上的显著差异。
正文 · 原文
Abstract:Forward-Forward learning replaces end-to-end backpropagation with local, layer-wise training objectives. Its variants are evaluated primarily by classification accuracy, often measured with linear probes trained on their representations. We show that this evaluation can over-credit the learning method: a linear probe on the representations of an untrained convolutional neural network already reaches accuracy well above chance. Moreover, a single accuracy value does not reveal which datapoints become correctly classified after training and which remain misclassified. To examine what learning contributes beyond the random, untrained baseline, we present an analysis methodology built on the Boundary Proximity Index (BPI), a multiclass margin normalized by a class-centroid reference. We apply our analysis to seven Forward-Forward variants and related layer-local methods and find that initially misclassified points near the probe's decision boundary generally have higher rescue rates than those farther inside incorrect decision regions. Rescue rates vary substantially across methods, with some correcting a substantial fraction of these deeply embedded points. Together, these analyses offer a way to evaluate representation learning beyond accuracy and identify the rescue of points far inside incorrect decision regions as a target for improving local learning methods.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.16347 [cs.LG] |
| (or arXiv:2610.16347v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16347 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Alireza Rahmansetayesh [view email]
[v1]
Mon, 12 Oct 2026 08:04:37 UTC (3,356 KB)
来源：arXiv cs.AI · arxiv.org