Reply to thread

Message

<blockquote data-quote="Trident" data-source="post: 1081689" data-attributes="member: 99014">Gradient boosted decision trees (XGBoost) are highly suitable to be ran on multi-core architecture, gpu or NPU locally, without cloud.Every leaf represents one feature with certain weight, based on how frequently it’s been seen in malicious or benign files.Left child will be entered (executed) when certain features are false (not found), right child&nbsp; will be entered when features are true (included). Gain for XGBoost is influenced by the count of the number of samples affected by the splits based on a feature.<img src="https://www.researchgate.net/publication/356698772/figure/fig2/AS:1096436418641951@1638422221975/The-architecture-of-Gradient-Boosting-Decision-Tree.png" alt="The architecture of Gradient Boosting Decision Tree | Download ..." class="fr-fic fr-dii fr-draggable " style="" /><img src="https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcSbWUl-pBLBpOHgHsI788TvNZTCpPiQcmKREk9fZJshcg&amp;s" alt="The architecture of Gradient Boosting Decision Tree | Download ..." class="fr-fic fr-dii fr-draggable " style="" />Example of how a local one looks like (static analysis, won’t mention vendor name).[SPOILER=&quot;Peek behind the tree encoding&quot;][ATTACH]282576[/ATTACH][ATTACH]282577[/ATTACH][ATTACH]282578[/ATTACH][ATTACH]282579[/ATTACH][/SPOILER]</blockquote>

[QUOTE="Trident, post: 1081689, member: 99014"] Gradient boosted decision trees (XGBoost) are highly suitable to be ran on multi-core architecture, gpu or NPU locally, without cloud. Every leaf represents one feature with certain weight, based on how frequently it’s been seen in malicious or benign files. Left child will be entered (executed) when certain features are false (not found), right child will be entered when features are true (included). Gain for XGBoost is influenced by the count of the number of samples affected by the splits based on a feature. [IMG alt="The architecture of Gradient Boosting Decision Tree | Download ..."]https://www.researchgate.net/publication/356698772/figure/fig2/AS:1096436418641951@1638422221975/The-architecture-of-Gradient-Boosting-Decision-Tree.png[/IMG][IMG alt="The architecture of Gradient Boosting Decision Tree | Download ..."]https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcSbWUl-pBLBpOHgHsI788TvNZTCpPiQcmKREk9fZJshcg&s[/IMG] Example of how a local one looks like (static analysis, won’t mention vendor name). [SPOILER="Peek behind the tree encoding"] [ATTACH alt="IMG_3330.jpeg"]282576[/ATTACH] [ATTACH alt="IMG_3331.jpeg"]282577[/ATTACH] [ATTACH alt="IMG_3332.jpeg"]282578[/ATTACH] [ATTACH alt="IMG_3333.jpeg"]282579[/ATTACH] [/SPOILER] [/QUOTE]

Verification