Unverified50% confidenceFactExact time
Upward substitution is the core operation when hierarchical questions are used for explanation: when the system needs to explain a deep question dimension, it searches upward along the hierarchy tree for an aligned shallow question and replaces it with a more general question, thereby reducing the number of questions that need to be presented to humans.
1
Sources
50%
Confidence
Long-term
Relevance
11/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedClassic Boosting algorithms (such as Gradient Boosting) can be understood as special cases of the functional gradient descent framework, where each iteration approximates the gradient direction with a new decision tree.60% similarUnverifiedThe core idea of matrix preconditioning is to apply a linear transformation to the gradient before the gradient update in order to improve the condition number.59% similarUnverifiedGradient boosted trees form an ensemble model by iteratively stacking multiple decision trees, each correcting the residual errors of the previous tree.58% similarUnverifiedThe hierarchical question structure is constructed through recursive clustering, contrastive prompting, and question alignment.58% similarVerifiedLayers at different depths in a Transformer have a division of labor: shallow layers handle lexical and syntactic processing, middle layers handle semantic integration, and deep layers handle higher-order reasoning and output distribution regulation.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1138217API
curl https://kongchang.com/api/v1/knowledge/claims/1138217MCP
get_claim(id=1138217)