This site's three trees weren't extracted from one published survey — they were sketched early in this project, before most of the matrix's references existed, synthesizing common ways the field organizes AI-based IDS work. As more papers were added since, several turned out to organize their own literature the same way. Those are listed below as precedent — evidence this taxonomy is consistent with the field, not evidence it was copied from anywhere.
Every reference cited by a method above (0 papers), numbered to match the badges on each method box. Click an entry for the full matrix record — venue, datasets, methods, metrics, contribution, gaps, and DOI.
Eight papers behind this review — five surveys, one empirical XAI framework paper, and two neuroscience-inspired ML papers grounding the DCOS direction — each run through an AI summarization pass that preserves the original authors' own structure and filters their citations to 2021–August 2026 — a companion to the numbered matrix above, not a replacement for reading the primary literature. Click a card to open the PDF.
A second tree, alongside the method taxonomy above: not what researchers built, but what they say is still blocking real-world deployment — and what they propose to do about it. Starts from Xu et al. (2025)'s discussion section (ref [2]) and grows the same way the method tree does, as more papers are added.
When each idea actually arrived — general AI/ML history and this matrix's IDS-specific literature, side by side, sorted by year rather than by taxonomy family. Two gaps become visible reading top to bottom: general techniques (the Transformer, then LLMs) take years to reach IDS specifically, and the context-gating/neuromodulation lane (Co4) never crosses into IDS work at all except one 2014 mobile-billing paper. The final entry names that second gap directly: DCOS, the synthesis this thesis proposes — not a cited result, but the space this timeline is built to make visible.
Small, focused implementations for building real intuition about each method — an appendix that grows alongside the thesis. Only methods with a written snippet appear below; click a card to open the real file in a new tab.
Not literature — programme requirements, milestones, and other reference documents worth keeping one click away. Click a card to open the real file in a new tab.