Unverified50% confidenceFactExact time
Neuroevolution uses evolutionary algorithms instead of gradient descent to optimize the weights or architecture of neural networks, requiring no differentiable loss function.
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
遗传算法+神经网络：登机效率超越Steffen法9.6%
Related Claims
Unverified等变神经网络的等变性使输出随输入的旋转或平移变换而以相同方式变换，这种归纳偏置提高了样本效率68% similarUnverified无梯度进化方法不要求目标函数可微，可优化包含离散操作、外部模拟器或硬件约束的复杂系统65% similarUnverifiedAnthropic提出'稀疏自编码器'（Sparse Autoencoder）方法试图将多义性神经元分解为更单纯的特征方向64% similarUnverifiedAnthropic's mechanistic interpretability research has been able to locate neural circuits associated with deceptive outputs in some cases63% similarUnverified卷积神经网络（CNN）和Transformer架构已被用于大脑皮层的自动分区62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/775777API
curl https://kongchang.com/api/v1/knowledge/claims/775777MCP
get_claim(id=775777)