Hot eventLive
论文:真实数据上Forward-Forward扩展性被合成基准高估
1 reports1 sources4 hr ago updated
Get the story
AI overview
2026年10月8日,arXiv Neural and Evolutionary Computing 发布论文(一手报道),指出现有合成基准高估了 Forward-Forward(FF)算法在真实数据上的扩展性,并提出 DTG-FF 方法,引入动态温度 goodness、解耦归一化与多层融合。论文在九个真实数据基准上达到 FF 家族当时最优:CIFAR-10 准确率 91.8%,并在 ImageNet-100 224×224 上建立 FF 基线。报道未涉及更早或相互矛盾的说法。
Generated from reports · updated 3 hr ago
Timeline
Follow the coverage from different angles.
Oct 8, 2026
- arXiv · Neural and Evolutionary Computing合成基准高估 Forward-Forward 扩展性:层局部训练的真实数据极限
论文提出 DTG-FF(动态温度 goodness、解耦归一化、多层融合),在九个真实数据基准上达到 FF 家族当时最优:CIFAR-10 91.8%,并在 ImageNet-100 224x224 建立 FF 基线。
Heat trend
Current heat 9·Comparable peak 10(Oct 8)·Comparable change over 24 hours –
The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.