analiza aiqbench
The authors have withdrawn their paper on timestamped QA for lecture videos due to measurement errors that invalidated their reported retrieval metrics and latency claims. This retraction underscores the critical importance of rigorous validation in cross-modal fusion research, particularly when optimizing for single-GPU constraints. For the tech community, the primary takeaway is the necessity to verify empirical results before drawing conclusions about model efficiency or performance gains.
CourseTimeQA: A Lecture-Video Benchmark and a Latency-Constrained Cross-Modal Fusion Method for Timestamped QA
arXiv:2512.00360v2 Announce Type: replace Abstract: We study timestamped question answering over educational lecture videos under a single-GPU latency/memory budget. Given a natural-language query, the system retrieves relevant timestamped segments and synthesizes a grounded answer. We present CourseTimeQA (52.3 h, 902 queries across six courses) and a lightweight, latency-constrained cross-modal retriever (CrossFusion-RAG) that combines frozen encoders, a learned 512->768 vision projection, shallow query-agnostic cross-attention over ASR and frames with a temporal-consistency regularizer, and a small cross-attentive reranker. On CourseTimeQA, CrossFusion-RAG improves nDCG@10 by 0.10 and MRR by 0.08 over a strong BLIP-2 retriever while achieving approximately 1.55 s median end-to-end latency on a single A100. Closest comparators (zero-shot CLIP multi-frame pooling; CLIP + cross-encoder reranker + MMR; learned late-fusion gating; text-only hybrid with cross-encoder reranking and its MMR variant; caption-augmented text retrieval; non-learned temporal smoothing) are evaluated under matched hardware and indexing. We report robustness across ASR noise (WER quartiles), diagnostics for temporal localization, and full training/tuning details to support reproducible comparison.
Abstract:We study timestamped question answering over educational lecture videos under a single-GPU latency/memory budget. Given a natural-language query, the system retrieves relevant timestamped segments and synthesizes a grounded answer. We present CourseTimeQA (52.3 h, 902 queries across six courses) and a lightweight, latency-constrained cross-modal retriever (CrossFusion-RAG) that combines frozen encoders, a learned 512->768 vision projection, shallow query-agnostic cross-attention over ASR and frames with a temporal-consistency regularizer, and a small cross-attentive reranker. On CourseTimeQA, CrossFusion-RAG improves nDCG@10 by 0.10 and MRR by 0.08 over a strong BLIP-2 retriever while achieving approximately 1.55 s median end-to-end latency on a single A100. Closest comparators (zero-shot CLIP multi-frame pooling; CLIP + cross-encoder reranker + MMR; learned late-fusion gating; text-only hybrid with cross-encoder reranking and its MMR variant; caption-augmented text retrieval; non-learned temporal smoothing) are evaluated under matched hardware and indexing. We report robustness across ASR noise (WER quartiles), diagnostics for temporal localization, and full training/tuning details to support reproducible comparison.
This paper is being withdrawn because an error in our measurement procedure produced incorrect values in our reported retrieval results (Tables I, II, V, and VI, and the corresponding headline figures in the Abstract). Several of our empirical claims depend on these measurements and therefore do not hold as stated
Computation and Language (cs.CL)
H.3.3; I.2.6; I.2.10
arXiv:2512.00360 [cs.CL]
(or arXiv:2512.00360v2 [cs.CL] for this version)
Text integral sub licența deschisă a publicației · 221 cuvinte
Subiecte conexe
[AINews] The Future of Latent Space
Latent Space · 25 sept. 2026
[AINews] Opus 5.5 is good at explainer videos
Latent Space · 29 sept. 2026
Build a Physical AI model factory with NVIDIA Cosmos 3 on SageMaker HyperPod
AWS · 4 sept. 2026
Beyond Independent Optimization: Compression, MoE Routing, and Quantization Interactions in Multimodal Edge Intelligence
arXiv · 24 iul. 2026