Unverified50% confidenceFactExact time
The lottery ticket hypothesis was proposed by Frankle and Carlin in 2019, pointing out that within a randomly initialized large neural network there exists a smaller subnetwork that, when trained in isolation, can achieve performance comparable to or even better than the original network.
1
Sources
50%
Confidence
Long-term
Relevance
10/8/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe universal approximation theorem for neural networks states that a sufficiently wide or sufficiently deep network can approximate any continuous function, but does not answer the question of what network size is required to achieve a specific accuracy.72% similarUnverified深度学习中的随机种子控制权重初始化、数据加载批次打乱顺序、Dropout神经元随机失活以及优化器中的随机梯度估计等多个随机性来源70% similarUnverifiedNNUE利用棋盘变化的稀疏性只更新受影响的网络激活值，推理速度比传统深度神经网络快两个数量级，可在CPU上实时运行68% similarUnverifiedNaïve post-training quantization causes individual weight errors to accumulate and amplify across multiple layers—a phenomenon known as compounding error—and at 2-bit or 1-bit precision the model structure collapses67% similarUnverified即便固定所有随机种子，在不同硬件或并行度下，训练结果仍可能出现被大模型放大为显著能力差异的细微偏差67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/988397API
curl https://kongchang.com/api/v1/knowledge/claims/988397MCP
get_claim(id=988397)