Abstract: This report synthesises findings from 3 peer-reviewed papers addressing the following research question: Can the adversarial contrastive learning framework be extended to improve robustness against adversarial attacks in cross-lingual rumor detection, as measured by accuracy on the MLQA benchmark. Only eighteen…

Abstract: This report synthesises findings from 2 peer-reviewed papers addressing the following research question: How does the adversarial contrastive learning approach compare to cross-lingual pre-training methods like mBERT in low-resource rumor detection accuracy on the XQuAD benchmark. Recent advancements in Large…

Abstract: This report synthesises findings from 4 peer-reviewed papers addressing the following research question: What is the computational efficiency trade-off between manifold-aware and Euclidean-based dense retrieval models when evaluating cross-lingual robustness on benchmarks like XLENT or mTEC. Cross-lingual…

Abstract: This report synthesises findings from 3 peer-reviewed papers addressing the following research question: Does the manifold-aware metric in MA-DPR improve zero-shot cross-lingual retrieval accuracy on benchmarks such as XQuAD compared to cosine similarity baselines. Cross-Lingual Retrieval Question Answering (CL-ReQA)…

Abstract: This report synthesises findings from 5 peer-reviewed papers addressing the following research question: To what extent does MA-DPR enhance robustness against adversarial query perturbations on domain-shifted datasets like NaturalQuestions compared to traditional dense retrieval models. Liang Wang, Nan Yang, Xiaolong…

Abstract: This report synthesises findings from 4 peer-reviewed papers addressing the following research question: How do recent multimodal dense retrieval models (e.g., M3-Rec) perform in misspelling robustness tasks compared to text-only models when evaluated on benchmarks like Flickr30k or XLENT with induced. Dialogue…

Abstract: This report synthesises findings from 13 peer-reviewed papers addressing the following research question: Does integrating manifold-aware fine-tuning improve the robustness of dense retrievers against adversarial perturbations in query embeddings compared to standard Euclidean-based models. In the last few years, the…

Abstract: This report synthesises findings from 6 peer-reviewed papers addressing the following research question: How does the performance of manifold-aware dense passage retrieval models compare to Euclidean-based models on out-of-domain misspelling robustness benchmarks like TypoBEIR or SpellingNoisyMSMARCO. In this paper…

Abstract: This report synthesises findings from 7 peer-reviewed papers addressing the following research question: How does manifold-aware distance metric integration affect zero-shot retrieval accuracy on BEIR OOD domains compared to standard cosine similarity in dense passage retrieval models. Decoder-only large language…

Abstract: This report synthesises findings from 13 peer-reviewed papers addressing the following research question: Does applying manifold-aware distance metrics improve cross-domain robustness in zero-shot retrieval for multimodal datasets compared to traditional DPR approaches. A FUNDAMENTAL CHALLENGE FOR SYSTEMS…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: How do dense retrieval methods optimized for robustness against text noise perform in terms of inference efficiency and latency compared to standard dual-encoder baselines on large-scale corpora. Abstract The…

Abstract: This report synthesises findings from 8 peer-reviewed papers addressing the following research question: What is the impact of input text noise on the retrieval precision of manifold-aware models versus standard dual-encoders when evaluated on out-of-domain NQ and TriviaQA benchmarks. Abstract Large language models…

Abstract: This report synthesises findings from 14 peer-reviewed papers addressing the following research question: Does manifold-aware embedding projection improve cross-domain robustness in recommendation-as-retrieval tasks compared to domain-adaptive fine-tuning alone. Clustering is the unsupervised classification of…

Abstract: This report synthesises findings from 12 peer-reviewed papers addressing the following research question: Does integrating manifold regularization into dense retrievers improve zero-shot cross-domain robustness on heterogeneous QA corpora compared to baseline dual-encoder models. This paper surveys the field of…

Abstract: This report synthesises findings from 14 peer-reviewed papers addressing the following research question: What is the impact of synthetic misspelling noise on the retrieval accuracy of dual-encoder architectures compared to contrastive learning methods on standard QA datasets. Language models (LMs) are becoming the…

Abstract: This report synthesises findings from 5 peer-reviewed papers addressing the following research question: How does the Recall@K of manifold-aware dense retrieval models compare to SimCLR baselines when evaluated on cross-domain OOD benchmarks transitioning from TriviaQA to HotpotQA.…

Abstract: This report synthesises findings from 10 peer-reviewed papers addressing the following research question: How does manifold-aware embedding projection compare to standard PCA in maintaining recall@K accuracy while reducing inference latency for billion-scale dense retrieval indexes. Abstract The rapid advances in the…

Abstract: This report synthesises findings from 5 peer-reviewed papers addressing the following research question: To what extent does fine-tuning mDPR with contrastive losses optimized for hyperbolic space improve zero-shot cross-lingual retrieval robustness on low-resource languages in the XOR-TyDi QA dataset. Human…

Abstract: This report synthesises findings from 6 peer-reviewed papers addressing the following research question: Does replacing linear similarity measures with non-linear manifold-based scoring in the CORA pipeline reduce the performance gap between high-resource and extremely low-resource languages on. This article provides…

Abstract: This report synthesises findings from 14 peer-reviewed papers addressing the following research question: Do manifold-aware embedding distances improve robustness against adversarial perturbations in out-of-distribution retrieval tasks across the BEIR dataset. Decoder-only large language models (LLMs) are…

Abstract: This report synthesises findings from 11 peer-reviewed papers addressing the following research question: How does integrating geodesic distance metrics into dense passage retrieval affect zero-shot performance on the BEIR benchmark compared to standard cosine similarity. In the last few years, the deep learning (DL)…

Abstract: This report synthesises findings from 5 peer-reviewed papers addressing the following research question: How does applying Riemannian manifold metrics to mDPR embeddings affect retrieval accuracy on the XOR-TyDi QA benchmark for Amharic and Kannada compared to standard Euclidean distance. Despite progress in…

Abstract: This report synthesises findings from 11 peer-reviewed papers addressing the following research question: What is the trade-off between inference latency and accuracy when deploying Llama-2 multimodal models for diagram-based code generation, as measured by pass@1 and throughput on HumanEval-V. In this report, we…

Abstract: This report synthesises findings from 15 peer-reviewed papers addressing the following research question: Does integrating manifold-aware distance metrics into the training objective improve the robustness of dense retrievers against adversarial perturbations in out-of-distribution query settings. Natural Language…

Abstract: This report synthesises findings from 6 peer-reviewed papers addressing the following research question: How does the performance of Llama-2-based multimodal models on diagram-to-code generation tasks vary with different image segmentation techniques, as measured by pass@1 and pass@k on HumanEval-V. The deployment of…