id
stringlengths
33
34
title
stringlengths
6
218
authors
stringlengths
5
6.98k
categories
stringlengths
5
188
summary
stringlengths
34
3.26k
link_pdf
stringlengths
33
34
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/2305.02996v2
Efficient k-NN Search with Cross-Encoders using Adaptive Multi-Round CUR Decomposition
Nishant Yadav, Nicholas Monath, Manzil Zaheer, Andrew McCallum
cs.IR, cs.CL, cs.LG
Cross-encoder models, which jointly encode and score a query-item pair, are prohibitively expensive for direct k-nearest neighbor (k-NN) search. Consequently, k-NN search typically employs a fast approximate retrieval (e.g. using BM25 or dual-encoder vectors), followed by reranking with a cross-encoder; however, the re...
http://arxiv.org/pdf/2305.02996v2
2023-10-23T17:48:34
2023-05-04T17:01:17
{ "title": [ -0.055908203, -0.0036621094, -0.036499023, -0.03829956, 0.0025024414, 0.024154663, -0.04147339, -0.004146576, 0.010650635, 0.052093506, -0.042144775, 0.021606445, 0.01576233, 0.005744934, -0.022155762, 0.009529114, -0.0016145706, 0.0...
http://arxiv.org/abs/2305.02995v2
Accuracy on the Curve: On the Nonlinear Correlation of ML Performance Between Data Subpopulations
Weixin Liang, Yining Mao, Yongchan Kwon, Xinyu Yang, James Zou
cs.LG, cs.AI, cs.CL, cs.CV
Understanding the performance of machine learning (ML) models across diverse data distributions is critically important for reliable applications. Despite recent empirical studies positing a near-perfect linear correlation between in-distribution (ID) and out-of-distribution (OOD) accuracies, we empirically demonstrate...
http://arxiv.org/pdf/2305.02995v2
2023-05-31T17:41:16
2023-05-04T17:00:17
{ "title": [ -0.016433716, 0.021636963, -0.04043579, -0.028427124, 0.028015137, -0.015106201, -0.009063721, -0.029647827, -0.014511108, 0.028564453, -0.030761719, 0.020111084, -0.028320312, -0.018371582, -0.0340271, -0.0015077591, -0.0041122437, ...
http://arxiv.org/abs/2305.02993v2
SemEval-2023 Task 7: Multi-Evidence Natural Language Inference for Clinical Trial Data
Maël Jullien, Marco Valentino, Hannah Frost, Paul O'Regan, Donal Landers, André Freitas
cs.CL, cs.AI, cs.LG
This paper describes the results of SemEval 2023 task 7 -- Multi-Evidence Natural Language Inference for Clinical Trial Data (NLI4CT) -- consisting of 2 tasks, a Natural Language Inference (NLI) task, and an evidence selection task on clinical trial data. The proposed challenges require multi-hop biomedical and numeric...
http://arxiv.org/pdf/2305.02993v2
2023-05-11T09:10:06
2023-05-04T16:58:19
{ "title": [ -0.026351929, -0.025772095, -0.05407715, -0.07159424, -0.011108398, 0.009315491, -0.071899414, 0.0074920654, -0.004371643, 0.021224976, -0.016708374, 0.017211914, -0.015151978, -0.018676758, -0.014915466, 0.011070251, -0.018859863, 0...
http://arxiv.org/abs/2305.03509v2
Diffusion Explainer: Visual Explanation for Text-to-image Stable Diffusion
Seongmin Lee, Benjamin Hoover, Hendrik Strobelt, Zijie J. Wang, ShengYun Peng, Austin Wright, Kevin Li, Haekyu Park, Haoyang Yang, Duen Horng Chau
cs.CL, cs.AI, cs.HC, cs.LG
Diffusion-based generative models' impressive ability to create convincing images has captured global attention. However, their complex internal structures and operations often make them difficult for non-experts to understand. We present Diffusion Explainer, the first interactive visualization tool that explains how S...
http://arxiv.org/pdf/2305.03509v2
2023-05-08T21:32:02
2023-05-04T16:14:43
{ "title": [ -0.0005030632, 0.0058555603, -0.020706177, -0.029266357, -0.013442993, -0.013214111, -0.018310547, -0.023132324, 0.0058288574, 0.030731201, -0.012512207, 0.010246277, 0.027038574, -0.039916992, -0.004131317, 0.0024929047, 0.007575989, ...
http://arxiv.org/abs/2305.06164v1
Conversational Semantic Parsing using Dynamic Context Graphs
Parag Jain, Mirella Lapata
cs.CL, cs.AI
In this paper we consider the task of conversational semantic parsing over general purpose knowledge graphs (KGs) with millions of entities, and thousands of relation-types. We are interested in developing models capable of interactively mapping user utterances into executable logical forms (e.g., SPARQL) in the contex...
http://arxiv.org/pdf/2305.06164v1
2023-05-04T16:04:41
2023-05-04T16:04:41
{ "title": [ -0.01689148, -0.0015172958, -0.042114258, -0.097595215, 0.009628296, 0.01725769, -0.07965088, 0.014549255, -0.003042221, 0.037139893, -0.02444458, 0.011886597, -0.01789856, 0.002313614, -0.053863525, -0.013595581, -0.015602112, 0.039...
http://arxiv.org/abs/2305.02937v2
End-to-end spoken language understanding using joint CTC loss and self-supervised, pretrained acoustic encoders
Jixuan Wang, Martin Radfar, Kai Wei, Clement Chung
cs.CL, cs.SD, eess.AS
It is challenging to extract semantic meanings directly from audio signals in spoken language understanding (SLU), due to the lack of textual information. Popular end-to-end (E2E) SLU models utilize sequence-to-sequence automatic speech recognition (ASR) models to extract textual embeddings as input to infer semantics,...
http://arxiv.org/pdf/2305.02937v2
2023-06-02T13:25:06
2023-05-04T15:36:37
{ "title": [ -0.023651123, -0.016098022, -0.021438599, -0.060760498, -0.02861023, 0.012451172, -0.062164307, 0.013900757, 0.0049591064, 0.028884888, -0.013023376, -0.0042495728, -0.0033912659, -0.021606445, -0.03050232, 0.007003784, -0.001824379, ...
http://arxiv.org/abs/2305.03495v2
Automatic Prompt Optimization with "Gradient Descent" and Beam Search
Reid Pryzant, Dan Iter, Jerry Li, Yin Tat Lee, Chenguang Zhu, Michael Zeng
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have shown impressive performance as general purpose agents, but their abilities remain highly dependent on prompts which are hand written with onerous trial-and-error effort. We propose a simple and nonparametric solution to this problem, Automatic Prompt Optimization (APO), which is inspi...
http://arxiv.org/pdf/2305.03495v2
2023-10-19T04:37:25
2023-05-04T15:15:22
{ "title": [ -0.029785156, -0.0025577545, -0.023239136, -0.027801514, -0.022323608, -0.00005620718, -0.058380127, -0.03756714, -0.016784668, 0.031311035, -0.007858276, -0.0036792755, -0.00046634674, -0.05053711, -0.017456055, 0.0005607605, 0.00946807...
http://arxiv.org/abs/2305.02897v2
An automatically discovered chain-of-thought prompt generalizes to novel models and datasets
Konstantin Hebenstreit, Robert Praas, Louis P Kiesewetter, Matthias Samwald
cs.CL, cs.AI
Emergent chain-of-thought (CoT) reasoning capabilities promise to improve performance and explainability of large language models (LLMs). However, uncertainties remain about how reasoning strategies formulated for previous model generations generalize to new model generations and different datasets. In this small-scale...
http://arxiv.org/pdf/2305.02897v2
2023-08-03T14:33:37
2023-05-04T15:07:20
{ "title": [ 0.04849243, 0.001865387, 0.0020751953, -0.022888184, -0.02507019, -0.023117065, -0.031707764, -0.029373169, 0.0017881393, 0.016067505, -0.053497314, 0.027374268, -0.010322571, -0.012649536, 0.014587402, -0.023468018, 0.022003174, 0.0...
http://arxiv.org/abs/2305.02869v1
2x Faster Language Model Pre-training via Masked Structural Growth
Yiqun Yao, Zheng Zhang, Jing Li, Yequan Wang
cs.CL
Acceleration of large language model pre-training is a critical issue in present NLP research. In this paper, we focus on speeding up pre-training by progressively growing from a small Transformer structure to a large one. There are two main research problems related to progressive growth: growth schedule and growth op...
http://arxiv.org/pdf/2305.02869v1
2023-05-04T14:28:39
2023-05-04T14:28:39
{ "title": [ -0.02557373, -0.013832092, -0.057800293, -0.06677246, -0.020874023, -0.011177063, -0.0690918, 0.018371582, -0.02470398, 0.032409668, -0.023834229, 0.023727417, 0.0178833, -0.029769897, 0.010467529, 0.0141067505, 0.0028743744, 0.03454...
http://arxiv.org/abs/2305.02865v1
CausalAPM: Generalizable Literal Disentanglement for NLU Debiasing
Songyang Gao, Shihan Dou, Junjie Shan, Qi Zhang, Xuanjing Huang
cs.CL, cs.AI
Dataset bias, i.e., the over-reliance on dataset-specific literal heuristics, is getting increasing attention for its detrimental effect on the generalization ability of NLU models. Existing works focus on eliminating dataset bias by down-weighting problematic data in the training process, which induce the omission of ...
http://arxiv.org/pdf/2305.02865v1
2023-05-04T14:22:26
2023-05-04T14:22:26
{ "title": [ -0.0154953, 0.0053596497, -0.045043945, -0.08239746, -0.008399963, 0.005039215, -0.06817627, -0.014472961, -0.010665894, 0.04031372, -0.008796692, 0.010818481, 0.047698975, -0.024734497, -0.0058898926, 0.013626099, -0.012275696, 0.03...
http://arxiv.org/abs/2305.02858v1
ReMask: A Robust Information-Masking Approach for Domain Counterfactual Generation
Pengfei Hong, Rishabh Bhardwaj, Navonil Majumdar, Somak Aditya, Soujanya Poria
cs.CL, cs.AI
Domain shift is a big challenge in NLP, thus, many approaches resort to learning domain-invariant features to mitigate the inference phase domain shift. Such methods, however, fail to leverage the domain-specific nuances relevant to the task at hand. To avoid such drawbacks, domain counterfactual generation aims to tra...
http://arxiv.org/pdf/2305.02858v1
2023-05-04T14:19:02
2023-05-04T14:19:02
{ "title": [ 0.000320673, 0.0072746277, -0.04547119, -0.027801514, -0.00806427, -0.023239136, -0.039215088, -0.040527344, 0.021392822, 0.025115967, -0.00070762634, 0.009460449, 0.005874634, -0.045715332, -0.03427124, 0.009048462, -0.0113220215, 0...
http://arxiv.org/abs/2305.03519v1
Leveraging BERT Language Model for Arabic Long Document Classification
Muhammad AL-Qurishi
cs.CL
Given the number of Arabic speakers worldwide and the notably large amount of content in the web today in some fields such as law, medicine, or even news, documents of considerable length are produced regularly. Classifying those documents using traditional learning models is often impractical since extended length of ...
http://arxiv.org/pdf/2305.03519v1
2023-05-04T13:56:32
2023-05-04T13:56:32
{ "title": [ -0.024887085, -0.01953125, -0.01727295, -0.050811768, -0.015396118, 0.0040512085, -0.04168701, 0.012062073, -0.002243042, 0.024398804, -0.01576233, -0.01638794, -0.0065574646, -0.023361206, -0.0473938, -0.018341064, -0.012802124, 0.0...
http://arxiv.org/abs/2305.02820v2
Semantic Space Grounded Weighted Decoding for Multi-Attribute Controllable Dialogue Generation
Zhiling Zhang, Mengyue Wu, Kenny Q. Zhu
cs.CL
Controlling chatbot utterance generation with multiple attributes such as personalities, emotions and dialogue acts is a practically useful but under-studied problem. We propose a novel framework called DASC that possesses strong controllability with a weighted decoding paradigm, while improving generation quality with...
http://arxiv.org/pdf/2305.02820v2
2023-11-05T10:35:02
2023-05-04T13:35:27
{ "title": [ -0.033355713, -0.017425537, -0.05996704, -0.060821533, -0.018692017, -0.003610611, -0.05834961, 0.020828247, 0.0052223206, 0.038330078, 0.01927185, 0.011955261, 0.011154175, -0.03475952, -0.045166016, 0.013465881, -0.020263672, 0.042...
http://arxiv.org/abs/2305.02810v1
Interpretable Sentence Representation with Variational Autoencoders and Attention
Ghazi Felhi
cs.CL, cs.LG
In this thesis, we develop methods to enhance the interpretability of recent representation learning techniques in natural language processing (NLP) while accounting for the unavailability of annotated data. We choose to leverage Variational Autoencoders (VAEs) due to their efficiency in relating observations to latent...
http://arxiv.org/pdf/2305.02810v1
2023-05-04T13:16:15
2023-05-04T13:16:15
{ "title": [ -0.0055389404, 0.012626648, -0.030960083, -0.06365967, 0.007598877, 0.007221222, -0.06124878, -0.013069153, 0.02458191, 0.03390503, -0.0113220215, -0.007980347, 0.0046081543, 0.00029230118, -0.061706543, 0.005844116, -0.010055542, 0....
http://arxiv.org/abs/2305.04798v2
Multi-grained Hypergraph Interest Modeling for Conversational Recommendation
Chenzhan Shang, Yupeng Hou, Wayne Xin Zhao, Yaliang Li, Jing Zhang
cs.IR, cs.AI, cs.CL
Conversational recommender system (CRS) interacts with users through multi-turn dialogues in natural language, which aims to provide high-quality recommendations for user's instant information need. Although great efforts have been made to develop effective CRS, most of them still focus on the contextual information fr...
http://arxiv.org/pdf/2305.04798v2
2023-10-26T15:52:30
2023-05-04T13:13:44
{ "title": [ 0.019638062, 0.010658264, -0.04824829, -0.052093506, 0.005859375, -0.00995636, -0.042663574, -0.004146576, 0.021026611, 0.03250122, -0.009292603, 0.009376526, 0.0075149536, -0.02017212, -0.050567627, 0.0015592575, -0.025558472, 0.029...
http://arxiv.org/abs/2305.02797v2
The Elephant in the Room: Analyzing the Presence of Big Tech in Natural Language Processing Research
Mohamed Abdalla, Jan Philip Wahle, Terry Ruas, Aurélie Névéol, Fanny Ducel, Saif M. Mohammad, Karën Fort
cs.CL
Recent advances in deep learning methods for natural language processing (NLP) have created new business opportunities and made NLP research critical for industry development. As one of the big players in the field of NLP, together with governments and universities, it is important to track the influence of industry on...
http://arxiv.org/pdf/2305.02797v2
2023-05-09T10:39:11
2023-05-04T12:57:18
{ "title": [ -0.016647339, -0.009536743, -0.050994873, -0.035186768, 0.007347107, -0.023864746, -0.0007867813, 0.003446579, -0.059051514, 0.019012451, -0.042785645, 0.02331543, -0.0063095093, -0.010093689, -0.013145447, -0.03048706, -0.01902771, ...
http://arxiv.org/abs/2305.02790v1
BranchNorm: Robustly Scaling Extremely Deep Transformers
Yijin Liu, Xianfeng Zeng, Fandong Meng, Jie Zhou
cs.LG, cs.CL
Recently, DeepNorm scales Transformers into extremely deep (i.e., 1000 layers) and reveals the promising potential of deep scaling. To stabilize the training of deep models, DeepNorm (Wang et al., 2022) attempts to constrain the model update to a constant value. Although applying such a constraint can benefit the early...
http://arxiv.org/pdf/2305.02790v1
2023-05-04T12:46:12
2023-05-04T12:46:12
{ "title": [ -0.0059165955, 0.008216858, -0.038726807, -0.08093262, 0.018295288, 0.00031375885, -0.04336548, -0.014793396, 0.005016327, 0.031311035, -0.025863647, 0.019699097, 0.012489319, -0.022064209, -0.031143188, 0.005783081, 0.007858276, 0.0...
http://arxiv.org/abs/2305.02777v2
Unified Model Learning for Various Neural Machine Translation
Yunlong Liang, Fandong Meng, Jinan Xu, Jiaan Wang, Yufeng Chen, Jie Zhou
cs.CL, cs.AI
Existing neural machine translation (NMT) studies mainly focus on developing dataset-specific models based on data from different tasks (e.g., document translation and chat translation). Although the dataset-specific models have achieved impressive performance, it is cumbersome as each dataset demands a model to be des...
http://arxiv.org/pdf/2305.02777v2
2023-05-18T11:53:14
2023-05-04T12:21:52
{ "title": [ -0.02015686, 0.0047569275, -0.015235901, -0.052856445, -0.0052833557, 0.006111145, -0.05770874, -0.012107849, -0.019119263, 0.0635376, 0.0017404556, 0.015449524, 0.020141602, -0.013938904, -0.037384033, 0.032562256, -0.0076408386, 0....
http://arxiv.org/abs/2305.02770v3
The Politics of Language Choice: How the Russian-Ukrainian War Influences Ukrainians' Language Use on Twitter
Daniel Racek, Brittany I. Davidson, Paul W. Thurner, Xiao Xiang Zhu, Göran Kauermann
cs.CY, cs.CL, stat.AP
The use of language is innately political and often a vehicle of cultural identity as well as the basis for nation building. Here, we examine language choice and tweeting activity of Ukrainian citizens based on more than 4 million geo-tagged tweets from over 62,000 users before and during the Russian-Ukrainian War, fro...
http://arxiv.org/pdf/2305.02770v3
2023-06-06T11:56:03
2023-05-04T12:15:54
{ "title": [ 0.024047852, 0.0028629303, -0.03515625, -0.04534912, 0.014968872, -0.023101807, 0.010734558, -0.000042021275, -0.07232666, 0.06072998, -0.031463623, 0.042999268, 0.017150879, 0.0050315857, -0.021911621, -0.019683838, -0.0009965897, 0...
http://arxiv.org/abs/2305.02763v1
VendorLink: An NLP approach for Identifying & Linking Vendor Migrants & Potential Aliases on Darknet Markets
Vageesh Saxena, Nils Rethmeier, Gijs Van Dijck, Gerasimos Spanakis
cs.CY, cs.CL, cs.CR, cs.LG
The anonymity on the Darknet allows vendors to stay undetected by using multiple vendor aliases or frequently migrating between markets. Consequently, illegal markets and their connections are challenging to uncover on the Darknet. To identify relationships between illegal markets and their vendors, we propose VendorLi...
http://arxiv.org/pdf/2305.02763v1
2023-05-04T12:04:33
2023-05-04T12:04:33
{ "title": [ 0.0069389343, 0.004283905, -0.056030273, -0.017486572, 0.0034103394, -0.02507019, 0.0053977966, -0.0063438416, 0.018997192, 0.026138306, -0.01763916, 0.010787964, 0.021133423, -0.029434204, -0.010856628, -0.00843811, -0.033721924, 0....
http://arxiv.org/abs/2305.02750v2
A Survey on Proactive Dialogue Systems: Problems, Methods, and Prospects
Yang Deng, Wenqiang Lei, Wai Lam, Tat-Seng Chua
cs.CL, cs.AI
Proactive dialogue systems, related to a wide range of real-world conversational applications, equip the conversational agent with the capability of leading the conversation direction towards achieving pre-defined targets or fulfilling certain goals from the system side. It is empowered by advanced techniques to progre...
http://arxiv.org/pdf/2305.02750v2
2023-05-09T11:34:24
2023-05-04T11:38:49
{ "title": [ -0.0072784424, -0.0000094771385, -0.034088135, -0.07702637, 0.037719727, 0.00070667267, -0.02331543, 0.00093364716, 0.0037345886, 0.019073486, 0.016845703, -0.009101868, -0.027114868, -0.038238525, -0.049194336, 0.0010976791, -0.00712585...
http://arxiv.org/abs/2305.02747v1
Unsupervised Dialogue Topic Segmentation with Topic-aware Utterance Representation
Haoyu Gao, Rui Wang, Ting-En Lin, Yuchuan Wu, Min Yang, Fei Huang, Yongbin Li
cs.CL, cs.AI
Dialogue Topic Segmentation (DTS) plays an essential role in a variety of dialogue modeling tasks. Previous DTS methods either focus on semantic similarity or dialogue coherence to assess topic similarity for unsupervised dialogue segmentation. However, the topic similarity cannot be fully identified via semantic simil...
http://arxiv.org/pdf/2305.02747v1
2023-05-04T11:35:23
2023-05-04T11:35:23
{ "title": [ -0.008003235, 0.0047950745, -0.035491943, -0.05697632, 0.0060806274, 0.0010156631, -0.048431396, 0.0016670227, 0.01789856, 0.0073928833, -0.005844116, 0.011886597, -0.010543823, -0.022964478, -0.025238037, 0.005455017, -0.022644043, ...
http://arxiv.org/abs/2305.02718v1
An Asynchronous Updating Reinforcement Learning Framework for Task-oriented Dialog System
Sai Zhang, Yuwei Hu, Xiaojie Wang, Caixia Yuan
cs.CL, cs.AI, cs.HC, 68T50, I.2.7
Reinforcement learning has been applied to train the dialog systems in many works. Previous approaches divide the dialog system into multiple modules including DST (dialog state tracking) and DP (dialog policy), and train these modules simultaneously. However, different modules influence each other during training. The...
http://arxiv.org/pdf/2305.02718v1
2023-05-04T10:39:17
2023-05-04T10:39:17
{ "title": [ 0.00059747696, 0.010017395, -0.032287598, -0.061431885, 0.017715454, -0.012756348, -0.028244019, 0.0016841888, -0.01637268, 0.045684814, -0.002571106, -0.008575439, -0.026138306, -0.0158844, -0.05065918, -0.030929565, -0.0025100708, ...
http://arxiv.org/abs/2305.03506v3
SI-LSTM: Speaker Hybrid Long-short Term Memory and Cross Modal Attention for Emotion Recognition in Conversation
Xingwei Liang, You Zou, Ruifeng Xu
cs.CL, cs.AI, cs.HC
Emotion Recognition in Conversation~(ERC) across modalities is of vital importance for a variety of applications, including intelligent healthcare, artificial intelligence for conversation, and opinion mining over chat history. The crux of ERC is to model both cross-modality and cross-time interactions throughout the c...
http://arxiv.org/pdf/2305.03506v3
2023-06-06T12:19:35
2023-05-04T10:13:15
{ "title": [ 0.01864624, 0.014701843, -0.027679443, -0.049987793, 0.003709793, -0.019897461, -0.039123535, -0.026885986, 0.024887085, 0.03237915, -0.025146484, -0.014793396, 0.030136108, -0.023498535, -0.04031372, 0.0019407272, -0.014442444, 0.02...
http://arxiv.org/abs/2305.02687v2
Big Data and Large Numbers. Interpreting Zipf's Law
Horia-Nicolai L. Teodorescu
physics.soc-ph, cs.CL, math.ST, stat.TH, 11, 62, G.2; G.3; I.2.7; C.2
It turns out that some empirical facts in Big Data are the effects of properties of large numbers. Zipf's law 'noise' is an example of such an artefact. We expose several properties of the power law distributions and of similar distribution that occur when the population is finite and the rank and counts of elements in...
http://arxiv.org/pdf/2305.02687v2
2023-05-07T18:21:19
2023-05-04T10:03:37
{ "title": [ -0.0143585205, -0.006149292, -0.04071045, -0.050994873, 0.0053749084, -0.02583313, 0.0022773743, 0.00982666, -0.005794525, 0.015731812, -0.04147339, 0.007762909, -0.03237915, -0.029434204, -0.014511108, -0.019104004, 0.0061302185, 0....
http://arxiv.org/abs/2305.02679v2
Neighboring Words Affect Human Interpretation of Saliency Explanations
Alon Jacovi, Hendrik Schuff, Heike Adel, Ngoc Thang Vu, Yoav Goldberg
cs.CL, cs.HC
Word-level saliency explanations ("heat maps over words") are often used to communicate feature-attribution in text-based models. Recent studies found that superficial factors such as word length can distort human interpretation of the communicated saliency scores. We conduct a user study to investigate how the marking...
http://arxiv.org/pdf/2305.02679v2
2023-05-06T12:22:19
2023-05-04T09:50:25
{ "title": [ 0.014274597, -0.013076782, -0.0020275116, -0.014587402, -0.0340271, -0.033081055, 0.012748718, -0.016296387, -0.008483887, 0.03314209, -0.045654297, -0.01991272, 0.017074585, -0.005256653, -0.034698486, -0.022705078, 0.011642456, 0.0...
http://arxiv.org/abs/2305.02665v1
Learning Language-Specific Layers for Multilingual Machine Translation
Telmo Pessoa Pires, Robin M. Schmidt, Yi-Hsiu Liao, Stephan Peitz
cs.CL, cs.AI
Multilingual Machine Translation promises to improve translation quality between non-English languages. This is advantageous for several reasons, namely lower latency (no need to translate twice), and reduced error cascades (e.g., avoiding losing gender and formality information when translating through English). On th...
http://arxiv.org/pdf/2305.02665v1
2023-05-04T09:18:05
2023-05-04T09:18:05
{ "title": [ -0.035491943, -0.009399414, -0.008323669, -0.051879883, -0.009300232, 0.021453857, -0.048797607, 0.010215759, -0.0044898987, 0.05206299, -0.0019817352, 0.0115737915, 0.022323608, -0.02078247, -0.035736084, 0.027038574, -0.0035533905, ...
http://arxiv.org/abs/2305.02637v2
Towards Weakly-Supervised Hate Speech Classification Across Datasets
Yiping Jin, Leo Wanner, Vishakha Laxman Kadam, Alexander Shvets
cs.CL, cs.AI, cs.CY
As pointed out by several scholars, current research on hate speech (HS) recognition is characterized by unsystematic data creation strategies and diverging annotation schemata. Subsequently, supervised-learning models tend to generalize poorly to datasets they were not trained on, and the performance of the models tra...
http://arxiv.org/pdf/2305.02637v2
2023-05-30T10:59:41
2023-05-04T08:15:40
{ "title": [ 0.025802612, 0.018188477, -0.0368042, -0.05621338, 0.029052734, -0.02432251, -0.008964539, -0.009506226, -0.002073288, 0.02822876, -0.028793335, 0.018157959, -0.0021858215, -0.0071754456, -0.039886475, -0.005001068, -0.012504578, 0.0...
http://arxiv.org/abs/2305.02633v1
Conformal Nucleus Sampling
Shauli Ravfogel, Yoav Goldberg, Jacob Goldberger
cs.CL, cs.LG
Language models generate text based on successively sampling the next word. A decoding procedure based on nucleus (top-$p$) sampling chooses from the smallest possible set of words whose cumulative probability exceeds the probability $p$. In this work, we assess whether a top-$p$ set is indeed aligned with its probabil...
http://arxiv.org/pdf/2305.02633v1
2023-05-04T08:11:57
2023-05-04T08:11:57
{ "title": [ -0.010528564, 0.013671875, -0.02935791, 0.0054626465, -0.048553467, -0.023544312, -0.04626465, -0.032440186, 0.002149582, 0.06451416, -0.029678345, 0.004295349, 0.03878784, -0.015388489, -0.0030403137, 0.018798828, 0.023239136, -0.01...
http://arxiv.org/abs/2305.02632v1
A framework for the emergence and analysis of language in social learning agents
Tobias J. Wieczorek, Tatjana Tchumatchenko, Carlos Wert Carvajal, Maximilian F. Eggl
cs.CL, cs.AI, cs.LG
Artificial neural networks (ANNs) are increasingly used as research models, but questions remain about their generalizability and representational invariance. Biological neural networks under social constraints evolved to enable communicable representations, demonstrating generalization capabilities. This study propose...
http://arxiv.org/pdf/2305.02632v1
2023-05-04T08:11:01
2023-05-04T08:11:01
{ "title": [ -0.03463745, -0.00033140182, -0.034942627, -0.015365601, 0.007633209, -0.0104522705, -0.0072402954, -0.008979797, -0.015350342, 0.032440186, -0.024139404, 0.021697998, -0.013519287, 0.0027160645, -0.03692627, -0.0069236755, -0.0128479, ...
http://arxiv.org/abs/2305.02615v2
How to Enhance Causal Discrimination of Utterances: A Case on Affective Reasoning
Hang Chen, Jing Luo, Xinyu Yang, Wenjing Zhu
cs.CL, cs.AI
Our investigation into the Affective Reasoning in Conversation (ARC) task highlights the challenge of causal discrimination. Almost all existing models, including large language models (LLMs), excel at capturing semantic correlations within utterance embeddings but fall short in determining the specific causal relation...
http://arxiv.org/pdf/2305.02615v2
2023-10-13T09:02:23
2023-05-04T07:45:49
{ "title": [ 0.027282715, -0.004600525, -0.03765869, -0.08508301, -0.00007802248, -0.020568848, -0.01625061, 0.017807007, 0.015487671, 0.0068588257, -0.026824951, 0.0032672882, 0.006061554, -0.034576416, -0.030212402, -0.0129470825, -0.01966858, ...
http://arxiv.org/abs/2305.02607v1
DN at SemEval-2023 Task 12: Low-Resource Language Text Classification via Multilingual Pretrained Language Model Fine-tuning
Daniil Homskiy, Narek Maloyan
cs.CL
In recent years, sentiment analysis has gained significant importance in natural language processing. However, most existing models and datasets for sentiment analysis are developed for high-resource languages, such as English and Chinese, leaving low-resource languages, particularly African languages, largely unexplor...
http://arxiv.org/pdf/2305.02607v1
2023-05-04T07:28:45
2023-05-04T07:28:45
{ "title": [ -0.006416321, -0.017562866, -0.029342651, -0.052459717, -0.012817383, -0.0011987686, -0.057647705, -0.011428833, -0.009887695, 0.04916382, -0.008590698, 0.026489258, -0.0047340393, -0.0022010803, -0.042236328, 0.0016641617, -0.019729614,...
http://arxiv.org/abs/2305.02606v2
Re$^3$Dial: Retrieve, Reorganize and Rescale Dialogue Corpus for Long-Turn Open-Domain Dialogue Pre-training
Jiaxin Wen, Hao Zhou, Jian Guan, Minlie Huang
cs.CL
Pre-training on large-scale open-domain dialogue data can substantially improve the performance of dialogue models. However, the pre-trained dialogue model's ability to utilize long-range context is limited due to the scarcity of long-turn dialogue sessions. Most dialogues in existing pre-training corpora contain fewer...
http://arxiv.org/pdf/2305.02606v2
2023-10-22T07:41:06
2023-05-04T07:28:23
{ "title": [ -0.0071372986, -0.016677856, -0.042236328, -0.07104492, -0.003648758, -0.00945282, -0.044006348, 0.020385742, -0.0030155182, 0.0357666, -0.020187378, 0.010574341, 0.012283325, -0.020935059, -0.013648987, -0.018035889, -0.019851685, 0...
http://arxiv.org/abs/2305.02579v1
From Statistical Methods to Deep Learning, Automatic Keyphrase Prediction: A Survey
Binbin Xie, Jia Song, Liangying Shao, Suhang Wu, Xiangpeng Wei, Baosong Yang, Huan Lin, Jun Xie, Jinsong Su
cs.CL, cs.AI
Keyphrase prediction aims to generate phrases (keyphrases) that highly summarizes a given document. Recently, researchers have conducted in-depth studies on this task from various perspectives. In this paper, we comprehensively summarize representative studies from the perspectives of dominant models, datasets and eval...
http://arxiv.org/pdf/2305.02579v1
2023-05-04T06:22:50
2023-05-04T06:22:50
{ "title": [ -0.038970947, -0.0022697449, -0.022537231, -0.03277588, -0.0074806213, 0.006286621, -0.03579712, -0.010307312, 0.0079193115, 0.053863525, -0.004020691, 0.013076782, 0.019515991, -0.008293152, -0.05166626, 0.022964478, -0.026550293, 0...
http://arxiv.org/abs/2305.02564v1
RetroMAE-2: Duplex Masked Auto-Encoder For Pre-Training Retrieval-Oriented Language Models
Shitao Xiao, Zheng Liu, Yingxia Shao, Zhao Cao
cs.CL
To better support information retrieval tasks such as web search and open-domain question answering, growing effort is made to develop retrieval-oriented language models, e.g., RetroMAE and many others. Most of the existing works focus on improving the semantic representation capability for the contextualized embedding...
http://arxiv.org/pdf/2305.02564v1
2023-05-04T05:37:22
2023-05-04T05:37:22
{ "title": [ -0.039642334, 0.010787964, -0.015808105, -0.039215088, -0.024780273, 0.009025574, -0.07543945, -0.016815186, -0.0066108704, 0.046203613, 0.010673523, 0.0087890625, 0.04763794, -0.047058105, 0.003396988, 0.03857422, -0.000600338, 0.01...
http://arxiv.org/abs/2305.02558v1
Analyzing Hong Kong's Legal Judgments from a Computational Linguistics point-of-view
Sankalok Sen
cs.CL, cs.IR
Analysis and extraction of useful information from legal judgments using computational linguistics was one of the earliest problems posed in the domain of information retrieval. Presently, several commercial vendors exist who automate such tasks. However, a crucial bottleneck arises in the form of exorbitant pricing an...
http://arxiv.org/pdf/2305.02558v1
2023-05-04T05:23:11
2023-05-04T05:23:11
{ "title": [ -0.020370483, -0.011886597, -0.035736084, -0.053100586, -0.025482178, -0.0065345764, -0.03515625, 0.023071289, 0.018325806, 0.026031494, -0.056488037, -0.012504578, 0.007019043, -0.009056091, -0.054840088, -0.025970459, -0.011207581, ...
http://arxiv.org/abs/2305.02556v1
Faithful Question Answering with Monte-Carlo Planning
Ruixin Hong, Hongming Zhang, Hong Zhao, Dong Yu, Changshui Zhang
cs.CL
Although large language models demonstrate remarkable question-answering performances, revealing the intermediate reasoning steps that the models faithfully follow remains challenging. In this paper, we propose FAME (FAithful question answering with MontE-carlo planning) to answer questions based on faithful reasoning ...
http://arxiv.org/pdf/2305.02556v1
2023-05-04T05:21:36
2023-05-04T05:21:36
{ "title": [ -0.003282547, 0.004802704, -0.020385742, -0.07421875, 0.00522995, 0.01184845, -0.03302002, -0.003435135, -0.010536194, 0.014221191, -0.024017334, 0.028808594, -0.030136108, -0.0087890625, -0.014434814, 0.0015850067, -0.0060691833, 0....
http://arxiv.org/abs/2305.02549v2
FormNetV2: Multimodal Graph Contrastive Learning for Form Document Information Extraction
Chen-Yu Lee, Chun-Liang Li, Hao Zhang, Timothy Dozat, Vincent Perot, Guolong Su, Xiang Zhang, Kihyuk Sohn, Nikolai Glushnev, Renshen Wang, Joshua Ainslie, Shangbang Long, Siyang Qin, Yasuhisa Fujii, Nan Hua, Tomas Pfister
cs.CL, cs.CV, cs.LG
The recent advent of self-supervised pre-training techniques has led to a surge in the use of multimodal learning in form document understanding. However, existing approaches that extend the mask language modeling to other modalities require careful multi-task tuning, complex reconstruction target designs, or additiona...
http://arxiv.org/pdf/2305.02549v2
2023-06-13T04:27:14
2023-05-04T05:02:04
{ "title": [ -0.027175903, -0.008071899, -0.044525146, -0.05609131, -0.007385254, -0.0121154785, -0.053955078, -0.0063171387, 0.04019165, 0.0061569214, -0.027328491, 0.0012197495, 0.00043725967, -0.026687622, -0.035614014, -0.00111866, -0.024734497, ...
http://arxiv.org/abs/2305.02547v2
PersonaLLM: Investigating the Ability of GPT-3.5 to Express Personality Traits and Gender Differences
Hang Jiang, Xiajie Zhang, Xubo Cao, Jad Kabbara
cs.CL, cs.AI, cs.HC
Despite the many use cases for large language models (LLMs) in the design of chatbots in various industries and the research showing the importance of personalizing chatbots to cater to different personality traits, little work has been done to evaluate whether the behaviors of personalized LLMs can reflect certain per...
http://arxiv.org/pdf/2305.02547v2
2023-05-18T18:56:20
2023-05-04T04:58:00
{ "title": [ 0.0026683807, 0.03488159, -0.038970947, -0.021774292, 0.013656616, -0.026672363, -0.026382446, -0.02116394, -0.032928467, 0.02810669, 0.0050621033, 0.0028858185, 0.0030231476, -0.03793335, -0.0032234192, 0.0072402954, -0.016845703, 0...
http://arxiv.org/abs/2305.02531v5
Can LLMs Capture Intertemporal Preferences?
Ali Goli, Amandeep Singh
cs.CL, cs.AI
We explore the viability of Large Language Models (LLMs), specifically OpenAI's GPT-3.5 and GPT-4, in emulating human survey respondents and eliciting preferences, with a focus on intertemporal choices. Leveraging the extensive literature on intertemporal discounting for benchmarking, we examine responses from LLMs acr...
http://arxiv.org/pdf/2305.02531v5
2023-10-25T22:31:50
2023-05-04T03:51:31
{ "title": [ 0.041656494, 0.02104187, -0.02041626, -0.013252258, -0.002275467, -0.031921387, 0.006752014, -0.029876709, 0.010498047, -0.00028443336, -0.06506348, -0.0031414032, -0.03189087, -0.020858765, 0.004699707, -0.045013428, -0.014320374, 0...
http://arxiv.org/abs/2305.02519v1
ANetQA: A Large-scale Benchmark for Fine-grained Compositional Reasoning over Untrimmed Videos
Zhou Yu, Lixiang Zheng, Zhou Zhao, Fei Wu, Jianping Fan, Kui Ren, Jun Yu
cs.CV, cs.CL
Building benchmarks to systemically analyze different capabilities of video question answering (VideoQA) models is challenging yet crucial. Existing benchmarks often use non-compositional simple questions and suffer from language biases, making it difficult to diagnose model weaknesses incisively. A recent benchmark AG...
http://arxiv.org/pdf/2305.02519v1
2023-05-04T03:04:59
2023-05-04T03:04:59
{ "title": [ 0.00077056885, -0.0028095245, -0.03289795, -0.09509277, 0.012565613, 0.016235352, -0.068237305, 0.012901306, 0.020141602, 0.0236969, -0.016281128, -0.006313324, -0.0056381226, -0.0236969, -0.033935547, -0.006362915, 0.0043525696, 0.0...
http://arxiv.org/abs/2305.02517v1
USTC-NELSLIP at SemEval-2023 Task 2: Statistical Construction and Dual Adaptation of Gazetteer for Multilingual Complex NER
Jun-Yu Ma, Jia-Chen Gu, Jiajun Qi, Zhen-Hua Ling, Quan Liu, Xiaoyi Zhao
cs.CL
This paper describes the system developed by the USTC-NELSLIP team for SemEval-2023 Task 2 Multilingual Complex Named Entity Recognition (MultiCoNER II). A method named Statistical Construction and Dual Adaptation of Gazetteer (SCDAG) is proposed for Multilingual Complex NER. The method first utilizes a statistics-base...
http://arxiv.org/pdf/2305.02517v1
2023-05-04T03:00:46
2023-05-04T03:00:46
{ "title": [ -0.008926392, -0.022003174, -0.004108429, -0.024124146, -0.021835327, -0.0022068024, -0.03756714, -0.02746582, -0.0014953613, 0.037750244, 0.007385254, -0.0059051514, 0.045532227, -0.05053711, 0.0058898926, 0.017074585, -0.023086548, ...
http://arxiv.org/abs/2305.02499v1
AutoML-GPT: Automatic Machine Learning with GPT
Shujian Zhang, Chengyue Gong, Lemeng Wu, Xingchao Liu, Mingyuan Zhou
cs.CL, cs.AI, cs.CV, cs.LG, stat.ML
AI tasks encompass a wide range of domains and fields. While numerous AI models have been designed for specific tasks and applications, they often require considerable human efforts in finding the right model architecture, optimization algorithm, and hyperparameters. Recent advances in large language models (LLMs) like...
http://arxiv.org/pdf/2305.02499v1
2023-05-04T02:09:43
2023-05-04T02:09:43
{ "title": [ -0.011291504, -0.010131836, -0.03527832, -0.060791016, 0.009841919, 0.0063438416, -0.052246094, -0.029312134, -0.002243042, 0.016021729, -0.011650085, 0.01158905, -0.0109939575, -0.033691406, -0.024627686, 0.013710022, -0.0060691833, ...
http://arxiv.org/abs/2305.02483v1
ChatGPT-steered Editing Instructor for Customization of Abstractive Summarization
Wen Xiao, Yujia Xie, Giuseppe Carenini, Pengcheng He
cs.CL
Tailoring outputs of large language models, such as ChatGPT, to specific user needs remains a challenge despite their impressive generation quality. In this paper, we propose a tri-agent generation pipeline consisting of a generator, an instructor, and an editor to enhance the customization of generated outputs. The ge...
http://arxiv.org/pdf/2305.02483v1
2023-05-04T01:12:35
2023-05-04T01:12:35
{ "title": [ 0.016616821, -0.015625, -0.01637268, -0.038024902, -0.06585693, 0.00712204, -0.058166504, -0.0004892349, -0.029205322, 0.054382324, -0.0076560974, 0.000048995018, 0.019622803, -0.024108887, 0.003326416, -0.020843506, -0.019897461, 0....
http://arxiv.org/abs/2305.03518v1
Black-box Prompt Tuning with Subspace Learning
Yuanhang Zheng, Zhixing Tan, Peng Li, Yang Liu
cs.CL, cs.AI
Black-box prompt tuning uses derivative-free optimization algorithms to learn prompts in low-dimensional subspaces instead of back-propagating through the network of Large Language Models (LLMs). Recent studies have found that black-box prompt tuning lacks versatility across tasks and LLMs, which we believe is related ...
http://arxiv.org/pdf/2305.03518v1
2023-05-04T01:04:25
2023-05-04T01:04:25
{ "title": [ -0.01751709, -0.0019159317, -0.024002075, -0.042510986, 0.0012998581, -0.011451721, -0.03353882, -0.02746582, -0.016311646, 0.0053977966, -0.052337646, 0.024124146, -0.025054932, -0.01737976, -0.013870239, -0.011016846, 0.0031852722, ...
http://arxiv.org/abs/2305.02471v1
Toward the Automated Construction of Probabilistic Knowledge Graphs for the Maritime Domain
Fatemeh Shiri, Teresa Wang, Shirui Pan, Xiaojun Chang, Yuan-Fang Li, Reza Haffari, Van Nguyen, Shuang Yu
cs.AI, cs.CL
International maritime crime is becoming increasingly sophisticated, often associated with wider criminal networks. Detecting maritime threats by means of fusing data purely related to physical movement (i.e., those generated by physical sensors, or hard data) is not sufficient. This has led to research and development...
http://arxiv.org/pdf/2305.02471v1
2023-05-04T00:24:30
2023-05-04T00:24:30
{ "title": [ -0.035491943, -0.0026569366, -0.05935669, -0.05682373, 0.0013217926, -0.00630188, -0.05444336, -0.014175415, -0.0041275024, 0.003435135, -0.040496826, 0.020065308, -0.030914307, -0.016036987, -0.016189575, -0.019104004, -0.008552551, ...
http://arxiv.org/abs/2305.02468v3
Task-Optimized Adapters for an End-to-End Task-Oriented Dialogue System
Namo Bang, Jeehyun Lee, Myoung-Wan Koo
cs.CL
Task-Oriented Dialogue (TOD) systems are designed to carry out specific tasks by tracking dialogue states and generating appropriate responses to help users achieve defined goals. Recently, end-to-end dialogue models pre-trained based on large datasets have shown promising performance in the conversational system. Howe...
http://arxiv.org/pdf/2305.02468v3
2023-05-31T06:18:28
2023-05-04T00:17:49
{ "title": [ -0.015991211, 0.00024747849, -0.042114258, -0.066467285, 0.017288208, -0.010009766, -0.030563354, 0.012619019, 0.0019340515, 0.04046631, 0.014198303, -0.029754639, 0.0030765533, -0.029052734, -0.057556152, -0.0008201599, 0.00017297268, ...
http://arxiv.org/abs/2305.02466v1
Cognitive Reframing of Negative Thoughts through Human-Language Model Interaction
Ashish Sharma, Kevin Rushton, Inna Wanyin Lin, David Wadden, Khendra G. Lucas, Adam S. Miner, Theresa Nguyen, Tim Althoff
cs.CL, cs.HC, cs.SI
A proven therapeutic technique to overcome negative thoughts is to replace them with a more hopeful "reframed thought." Although therapy can help people practice and learn this Cognitive Reframing of Negative Thoughts, clinician shortages and mental health stigma commonly limit people's access to therapy. In this paper...
http://arxiv.org/pdf/2305.02466v1
2023-05-04T00:12:52
2023-05-04T00:12:52
{ "title": [ -0.0058631897, -0.015510559, -0.011100769, -0.02973938, -0.01826477, -0.009063721, -0.03753662, -0.035247803, 0.024307251, 0.035125732, -0.015213013, -0.008102417, 0.020462036, -0.035583496, -0.05267334, 0.005870819, 0.00002515316, 0...
http://arxiv.org/abs/2305.03517v2
Few-shot Domain-Adaptive Visually-fused Event Detection from Text
Farhad Moghimifar, Fatemeh Shiri, Van Nguyen, Reza Haffari, Yuan-Fang Li
cs.CL, cs.AI
Incorporating auxiliary modalities such as images into event detection models has attracted increasing interest over the last few years. The complexity of natural language in describing situations has motivated researchers to leverage the related visual context to improve event detection performance. However, current a...
http://arxiv.org/pdf/2305.03517v2
2023-06-05T00:58:41
2023-05-04T00:10:57
{ "title": [ 0.0017528534, -0.0061836243, -0.03250122, -0.06072998, 0.010955811, -0.006996155, -0.016815186, -0.011894226, 0.015075684, 0.0262146, -0.01247406, 0.019210815, 0.019241333, -0.01209259, -0.038269043, 0.0070533752, -0.004096985, 0.021...
http://arxiv.org/abs/2305.02459v2
Transfer and Active Learning for Dissonance Detection: Addressing the Rare-Class Challenge
Vasudha Varadarajan, Swanie Juhng, Syeda Mahwish, Xiaoran Liu, Jonah Luby, Christian Luhmann, H. Andrew Schwartz
cs.CL, cs.LG
While transformer-based systems have enabled greater accuracies with fewer training examples, data acquisition obstacles still persist for rare-class tasks -- when the class label is very infrequent (e.g. < 5% of samples). Active learning has in general been proposed to alleviate such challenges, but choice of selectio...
http://arxiv.org/pdf/2305.02459v2
2023-05-05T00:56:19
2023-05-03T23:29:05
{ "title": [ 0.031051636, 0.018814087, -0.016326904, -0.027893066, 0.036834717, -0.03942871, 0.013656616, -0.04031372, 0.033325195, -0.0011234283, -0.022033691, -0.008079529, 0.0017356873, -0.048858643, -0.022491455, -0.003129959, -0.01828003, 0....
http://arxiv.org/abs/2305.02457v1
Quantifying the Dissimilarity of Texts
Benjamin Shade, Eduardo G. Altmann
cs.CL, cond-mat.stat-mech, physics.soc-ph
Quantifying the dissimilarity of two texts is an important aspect of a number of natural language processing tasks, including semantic information retrieval, topic classification, and document clustering. In this paper, we compared the properties and performance of different dissimilarity measures $D$ using three diffe...
http://arxiv.org/pdf/2305.02457v1
2023-05-03T22:59:08
2023-05-03T22:59:08
{ "title": [ 0.0030727386, 0.03439331, -0.034362793, -0.030181885, -0.0013532639, -0.027038574, -0.03186035, -0.038208008, 0.020828247, 0.034942627, -0.056243896, 0.016235352, 0.01171875, -0.012504578, -0.016464233, -0.008293152, 0.012992859, 0.0...
http://arxiv.org/abs/2305.02437v2
Lift Yourself Up: Retrieval-augmented Text Generation with Self Memory
Xin Cheng, Di Luo, Xiuying Chen, Lemao Liu, Dongyan Zhao, Rui Yan
cs.CL, cs.AI
With direct access to human-written reference as memory, retrieval-augmented generation has achieved much progress in a wide range of text generation tasks. Since better memory would typically prompt better generation~(we define this as primal problem). The traditional approach for memory retrieval involves selecting m...
http://arxiv.org/pdf/2305.02437v2
2023-05-17T07:33:11
2023-05-03T21:40:54
{ "title": [ -0.015289307, -0.010879517, -0.032318115, -0.041931152, -0.008529663, -0.00085544586, -0.018371582, 0.008743286, 0.002620697, 0.0546875, 0.006652832, 0.02319336, 0.0012130737, -0.0041885376, -0.070129395, 0.021453857, -0.0067825317, ...
http://arxiv.org/abs/2305.02426v1
evaluating bert and parsbert for analyzing persian advertisement data
Ali Mehrban, Pegah Ahadian
cs.CL
This paper discusses the impact of the Internet on modern trading and the importance of data generated from these transactions for organizations to improve their marketing efforts. The paper uses the example of Divar, an online marketplace for buying and selling products and services in Iran, and presents a competition...
http://arxiv.org/pdf/2305.02426v1
2023-05-03T20:50:05
2023-05-03T20:50:05
{ "title": [ 0.03366089, -0.0128479, 0.013465881, -0.037475586, -0.0057868958, -0.009048462, -0.0073242188, -0.022140503, -0.03137207, 0.017486572, -0.044921875, 0.0046195984, -0.016693115, -0.026321411, -0.008628845, -0.026992798, -0.0234375, 0....
http://arxiv.org/abs/2305.02424v1
Backdoor Learning on Sequence to Sequence Models
Lichang Chen, Minhao Cheng, Heng Huang
cs.CL
Backdoor learning has become an emerging research area towards building a trustworthy machine learning system. While a lot of works have studied the hidden danger of backdoor attacks in image or text classification, there is a limited understanding of the model's robustness on backdoor attacks when the output space is ...
http://arxiv.org/pdf/2305.02424v1
2023-05-03T20:31:13
2023-05-03T20:31:13
{ "title": [ -0.025512695, 0.00035524368, -0.026641846, -0.06512451, 0.0022029877, 0.011695862, -0.040863037, -0.009178162, -0.013130188, 0.028701782, -0.005207062, 0.011199951, -0.020248413, -0.028121948, -0.031204224, 0.0026435852, 0.0025100708, ...
http://arxiv.org/abs/2305.02423v1
PTP: Boosting Stability and Performance of Prompt Tuning with Perturbation-Based Regularizer
Lichang Chen, Heng Huang, Minhao Cheng
cs.CL, cs.AI
Recent studies show that prompt tuning can better leverage the power of large language models than fine-tuning on downstream natural language understanding tasks. However, the existing prompt tuning methods have training instability issues, as the variance of scores under different random seeds is quite large. To addre...
http://arxiv.org/pdf/2305.02423v1
2023-05-03T20:30:51
2023-05-03T20:30:51
{ "title": [ -0.014823914, -0.0038108826, -0.03152466, -0.07867432, 0.007888794, -0.0033817291, -0.06213379, -0.0076293945, -0.019714355, 0.00074386597, -0.008850098, 0.026382446, -0.009742737, -0.052642822, 0.0079422, 0.0066070557, 0.0027656555, ...
http://arxiv.org/abs/2305.02412v2
Plan, Eliminate, and Track -- Language Models are Good Teachers for Embodied Agents
Yue Wu, So Yeon Min, Yonatan Bisk, Ruslan Salakhutdinov, Amos Azaria, Yuanzhi Li, Tom Mitchell, Shrimai Prabhumoye
cs.CL, cs.AI, cs.LG
Pre-trained large language models (LLMs) capture procedural knowledge about the world. Recent work has leveraged LLM's ability to generate abstract plans to simplify challenging control tasks, either by action scoring, or action modeling (fine-tuning). However, the transformer architecture inherits several constraints ...
http://arxiv.org/pdf/2305.02412v2
2023-05-07T05:33:10
2023-05-03T20:11:22
{ "title": [ -0.012611389, -0.0087890625, -0.0005736351, -0.01625061, -0.018005371, -0.009712219, -0.023895264, -0.027557373, -0.012809753, 0.016906738, -0.023742676, 0.009742737, 0.009788513, -0.026687622, -0.0051193237, -0.002073288, -0.0070877075,...
http://arxiv.org/abs/2305.03513v2
ChatGraph: Interpretable Text Classification by Converting ChatGPT Knowledge to Graphs
Yucheng Shi, Hehuan Ma, Wenliang Zhong, Qiaoyu Tan, Gengchen Mai, Xiang Li, Tianming Liu, Junzhou Huang
cs.CL, cs.AI, cs.LG
ChatGPT, as a recently launched large language model (LLM), has shown superior performance in various natural language processing (NLP) tasks. However, two major limitations hinder its potential applications: (1) the inflexibility of finetuning on downstream tasks and (2) the lack of interpretability in the decision-ma...
http://arxiv.org/pdf/2305.03513v2
2023-09-19T14:26:17
2023-05-03T19:57:43
{ "title": [ -0.027893066, -0.009643555, -0.035003662, -0.04031372, -0.007736206, 0.009468079, -0.03463745, -0.012329102, -0.0018415451, 0.008598328, -0.022155762, 0.022537231, -0.031921387, -0.0067596436, -0.03866577, -0.015808105, -0.028366089, ...
http://arxiv.org/abs/2305.02394v2
Defending against Insertion-based Textual Backdoor Attacks via Attribution
Jiazhao Li, Zhuofeng Wu, Wei Ping, Chaowei Xiao, V. G. Vinod Vydiswaran
cs.CL, cs.AI, cs.CR, cs.LG
Textual backdoor attack, as a novel attack model, has been shown to be effective in adding a backdoor to the model during training. Defending against such backdoor attacks has become urgent and important. In this paper, we propose AttDef, an efficient attribution-based pipeline to defend against two insertion-based poi...
http://arxiv.org/pdf/2305.02394v2
2023-08-07T03:07:59
2023-05-03T19:29:26
{ "title": [ -0.0049476624, -0.004386902, -0.018249512, -0.05758667, 0.019439697, 0.009567261, -0.022827148, -0.016113281, -0.06530762, 0.030883789, 0.014251709, 0.022491455, -0.01737976, -0.027923584, -0.020477295, -0.012779236, -0.022750854, 0....
http://arxiv.org/abs/2305.02386v2
Approximating CKY with Transformers
Ghazal Khalighinejad, Ollie Liu, Sam Wiseman
cs.CL, cs.LG
We investigate the ability of transformer models to approximate the CKY algorithm, using them to directly predict a sentence's parse and thus avoid the CKY algorithm's cubic dependence on sentence length. We find that on standard constituency parsing benchmarks this approach achieves competitive or better performance t...
http://arxiv.org/pdf/2305.02386v2
2023-11-05T00:30:31
2023-05-03T18:55:09
{ "title": [ -0.016113281, 0.019836426, -0.0020427704, -0.038482666, -0.023742676, 0.031555176, -0.047790527, -0.034118652, -0.017852783, 0.07476807, -0.01789856, 0.003665924, 0.052246094, -0.030807495, -0.012519836, 0.026657104, 0.025314331, 0.0...
http://arxiv.org/abs/2305.02374v1
A Novel Plagiarism Detection Approach Combining BERT-based Word Embedding, Attention-based LSTMs and an Improved Differential Evolution Algorithm
Seyed Vahid Moravvej, Seyed Jalaleddin Mousavirad, Diego Oliva, Fardin Mohammadi
cs.CL, cs.LG, cs.NE
Detecting plagiarism involves finding similar items in two different sources. In this article, we propose a novel method for detecting plagiarism that is based on attention mechanism-based long short-term memory (LSTM) and bidirectional encoder representations from transformers (BERT) word embedding, enhanced with opti...
http://arxiv.org/pdf/2305.02374v1
2023-05-03T18:26:47
2023-05-03T18:26:47
{ "title": [ 0.01121521, 0.010917664, -0.004337311, -0.006866455, -0.007133484, -0.019088745, -0.0028495789, -0.045196533, -0.0077438354, 0.042388916, -0.013656616, 0.020980835, 0.019104004, -0.016586304, -0.025299072, 0.005470276, -0.030410767, ...
http://arxiv.org/abs/2305.02364v2
PeaCoK: Persona Commonsense Knowledge for Consistent and Engaging Narratives
Silin Gao, Beatriz Borges, Soyoung Oh, Deniz Bayazit, Saya Kanno, Hiromi Wakaki, Yuki Mitsufuji, Antoine Bosselut
cs.CL
Sustaining coherent and engaging narratives requires dialogue or storytelling agents to understand how the personas of speakers or listeners ground the narrative. Specifically, these agents must infer personas of their listeners to produce statements that cater to their interests. They must also learn to maintain consi...
http://arxiv.org/pdf/2305.02364v2
2023-05-26T08:45:23
2023-05-03T18:02:22
{ "title": [ -0.032165527, -0.0418396, -0.04248047, -0.053588867, -0.022354126, 0.0012321472, -0.028884888, 0.012290955, 0.0029468536, 0.032104492, -0.00945282, 0.006023407, 0.007205963, -0.023025513, -0.03366089, 0.0035152435, -0.026062012, 0.04...
http://arxiv.org/abs/2305.02363v2
Entity Tracking in Language Models
Najoung Kim, Sebastian Schuster
cs.CL
Keeping track of how states of entities change as a text or dialog unfolds is a key prerequisite to discourse understanding. Yet, there have been few systematic investigations into the ability of large language models (LLMs) to track discourse entities. In this work, we present a task probing to what extent a language ...
http://arxiv.org/pdf/2305.02363v2
2023-09-08T17:51:51
2023-05-03T18:01:13
{ "title": [ -0.011306763, -0.008552551, -0.03805542, -0.08319092, 0.0036888123, -0.00040960312, -0.05355835, 0.006866455, -0.00053834915, 0.040130615, 0.0010910034, -0.004180908, 0.046813965, -0.02810669, -0.028152466, 0.032318115, -0.010101318, ...
http://arxiv.org/abs/2305.02350v2
Using Language Models on Low-end Hardware
Fabian Ziegner, Janos Borst, Andreas Niekler, Martin Potthast
cs.CL, cs.LG
This paper evaluates the viability of using fixed language models for training text classification networks on low-end hardware. We combine language models with a CNN architecture and put together a comprehensive benchmark with 8 datasets covering single-label and multi-label classification of topic, sentiment, and gen...
http://arxiv.org/pdf/2305.02350v2
2023-05-08T15:43:52
2023-05-03T18:00:03
{ "title": [ -0.024597168, -0.017349243, -0.019042969, -0.02758789, -0.07128906, 0.0012311935, -0.050933838, 0.0129776, 0.016311646, 0.05618286, -0.028366089, -0.046722412, 0.034179688, -0.034332275, -0.041656494, 0.012069702, 0.026306152, 0.0225...
http://arxiv.org/abs/2305.02321v2
Entity-Based Evaluation of Political Bias in Automatic Summarization
Karen Zhou, Chenhao Tan
cs.CL, cs.AI, cs.CY
Growing literature has shown that NLP systems may encode social biases; however, the political bias of summarization models remains relatively unknown. In this work, we use an entity replacement method to investigate the portrayal of politicians in automatically generated summaries of news articles. We develop an entit...
http://arxiv.org/pdf/2305.02321v2
2023-10-19T18:15:34
2023-05-03T17:59:59
{ "title": [ 0.01763916, 0.017807007, -0.029403687, -0.05697632, 0.015205383, -0.021118164, -0.015945435, -0.028366089, -0.017074585, 0.04586792, -0.013870239, 0.019714355, 0.0023860931, 0.009475708, -0.053771973, -0.0026626587, -0.028915405, 0.0...
http://arxiv.org/abs/2305.02317v2
Visual Chain of Thought: Bridging Logical Gaps with Multimodal Infillings
Daniel Rose, Vaishnavi Himakunthala, Andy Ouyang, Ryan He, Alex Mei, Yujie Lu, Michael Saxon, Chinmay Sonar, Diba Mirza, William Yang Wang
cs.CL, cs.CV
Recent advances in large language models elicit reasoning in a chain of thought that allows models to decompose problems in a human-like fashion. Though this paradigm improves multi-step reasoning ability in language models, it is limited by being unimodal and applied mainly to question-answering tasks. We claim that i...
http://arxiv.org/pdf/2305.02317v2
2023-11-10T06:19:36
2023-05-03T17:58:29
{ "title": [ 0.007270813, -0.009429932, -0.02696228, -0.051727295, 0.002571106, -0.011291504, -0.023376465, -0.02734375, 0.013717651, 0.018249512, -0.04159546, -0.011207581, -0.0102005005, -0.023330688, -0.03857422, -0.034362793, -0.0011062622, 0...
http://arxiv.org/abs/2305.02301v2
Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizes
Cheng-Yu Hsieh, Chun-Liang Li, Chih-Kuan Yeh, Hootan Nakhost, Yasuhisa Fujii, Alexander Ratner, Ranjay Krishna, Chen-Yu Lee, Tomas Pfister
cs.CL, cs.AI, cs.LG
Deploying large language models (LLMs) is challenging because they are memory inefficient and compute-intensive for practical applications. In reaction, researchers train smaller task-specific models by either finetuning with human labels or distilling using LLM-generated labels. However, finetuning and distillation re...
http://arxiv.org/pdf/2305.02301v2
2023-07-05T16:59:31
2023-05-03T17:50:56
{ "title": [ -0.008934021, 0.002691269, -0.07409668, -0.060180664, -0.02734375, -0.026809692, -0.07348633, 0.0006508827, -0.004173279, 0.019348145, -0.05105591, 0.013206482, -0.0007529259, -0.030029297, -0.0018482208, -0.0340271, 0.015419006, 0.0...
http://arxiv.org/abs/2305.02300v1
Evaluating the Efficacy of Length-Controllable Machine Translation
Hao Cheng, Meng Zhang, Weixuan Wang, Liangyou Li, Qun Liu, Zhihua Zhang
cs.CL, cs.AI
Length-controllable machine translation is a type of constrained translation. It aims to contain the original meaning as much as possible while controlling the length of the translation. We can use automatic summarization or machine translation evaluation metrics for length-controllable machine translation, but this is...
http://arxiv.org/pdf/2305.02300v1
2023-05-03T17:50:33
2023-05-03T17:50:33
{ "title": [ -0.046051025, -0.01461792, -0.015403748, -0.023757935, -0.030212402, 0.005744934, -0.039154053, -0.0067100525, -0.025970459, 0.052825928, -0.0021400452, 0.02520752, 0.013313293, -0.017196655, -0.037841797, 0.023101807, -0.003917694, ...
http://arxiv.org/abs/2305.02291v1
Evaluating BERT-based Scientific Relation Classifiers for Scholarly Knowledge Graph Construction on Digital Library Collections
Ming Jiang, Jennifer D'Souza, Sören Auer, J. Stephen Downie
cs.DL, cs.CL
The rapid growth of research publications has placed great demands on digital libraries (DL) for advanced information management technologies. To cater to these demands, techniques relying on knowledge-graph structures are being advocated. In such graph-based pipelines, inferring semantic relations between related scie...
http://arxiv.org/pdf/2305.02291v1
2023-05-03T17:32:16
2023-05-03T17:32:16
{ "title": [ -0.012817383, -0.0060157776, -0.037078857, -0.024429321, -0.015792847, -0.013999939, -0.015548706, -0.0075263977, 0.014282227, 0.021957397, 0.0035209656, -0.000582695, -0.013793945, -0.023712158, -0.060028076, -0.018295288, -0.012580872,...
http://arxiv.org/abs/2305.02269v1
M2-CTTS: End-to-End Multi-scale Multi-modal Conversational Text-to-Speech Synthesis
Jinlong Xue, Yayue Deng, Fengping Wang, Ya Li, Yingming Gao, Jianhua Tao, Jianqing Sun, Jiaen Liang
cs.SD, cs.CL, eess.AS
Conversational text-to-speech (TTS) aims to synthesize speech with proper prosody of reply based on the historical conversation. However, it is still a challenge to comprehensively model the conversation, and a majority of conversational TTS systems only focus on extracting global information and omit local prosody fea...
http://arxiv.org/pdf/2305.02269v1
2023-05-03T16:59:38
2023-05-03T16:59:38
{ "title": [ 0.01234436, -0.008361816, -0.037475586, -0.0713501, -0.016998291, -0.000032305717, -0.0713501, 0.010856628, -0.0057258606, 0.060699463, -0.03463745, 0.025375366, 0.06842041, -0.019577026, -0.019943237, 0.0107803345, -0.014724731, 0.0...
http://arxiv.org/abs/2305.02265v2
A Neural Divide-and-Conquer Reasoning Framework for Image Retrieval from Linguistically Complex Text
Yunxin Li, Baotian Hu, Yuxin Ding, Lin Ma, Min Zhang
cs.CL
Pretrained Vision-Language Models (VLMs) have achieved remarkable performance in image retrieval from text. However, their performance drops drastically when confronted with linguistically complex texts that they struggle to comprehend. Inspired by the Divide-and-Conquer algorithm and dual-process theory, in this paper...
http://arxiv.org/pdf/2305.02265v2
2023-05-05T17:41:57
2023-05-03T16:55:00
{ "title": [ 0.018112183, 0.0034370422, 0.018447876, -0.03286743, 0.0060920715, 0.0028514862, 0.0063819885, -0.022262573, 0.028442383, 0.050842285, -0.043518066, 0.019897461, 0.0037136078, 0.014213562, -0.058135986, 0.0024528503, 0.00945282, -0.0...
http://arxiv.org/abs/2305.02261v1
End-to-end Training and Decoding for Pivot-based Cascaded Translation Model
Hao Cheng, Meng Zhang, Liangyou Li, Qun Liu, Zhihua Zhang
cs.CL, cs.AI
Utilizing pivot language effectively can significantly improve low-resource machine translation. Usually, the two translation models, source-pivot and pivot-target, are trained individually and do not utilize the limited (source, target) parallel data. This work proposes an end-to-end training method for the cascaded t...
http://arxiv.org/pdf/2305.02261v1
2023-05-03T16:48:43
2023-05-03T16:48:43
{ "title": [ -0.04626465, -0.007167816, 0.004890442, -0.064453125, -0.03173828, 0.034301758, -0.07305908, -0.017852783, -0.0017328262, 0.06536865, -0.0061149597, -0.007888794, 0.039245605, -0.031707764, -0.041015625, 0.019134521, 0.0035057068, 0....
http://arxiv.org/abs/2305.02239v2
The Benefits of Label-Description Training for Zero-Shot Text Classification
Lingyu Gao, Debanjan Ghosh, Kevin Gimpel
cs.CL, cs.AI
Pretrained language models have improved zero-shot text classification by allowing the transfer of semantic knowledge from the training data in order to classify among specific label sets in downstream tasks. We propose a simple way to further improve zero-shot accuracies with minimal effort. We curate small finetuning...
http://arxiv.org/pdf/2305.02239v2
2023-10-23T15:24:57
2023-05-03T16:19:31
{ "title": [ -0.0034599304, -0.0042266846, -0.040863037, -0.041870117, -0.0015478134, -0.024276733, -0.008888245, 0.0019893646, -0.00894928, 0.04196167, -0.006122589, 0.016662598, -0.009902954, -0.015052795, -0.04977417, -0.00315094, -0.012458801, ...
http://arxiv.org/abs/2305.02235v1
AttenWalker: Unsupervised Long-Document Question Answering via Attention-based Graph Walking
Yuxiang Nie, Heyan Huang, Wei Wei, Xian-Ling Mao
cs.CL
Annotating long-document question answering (long-document QA) pairs is time-consuming and expensive. To alleviate the problem, it might be possible to generate long-document QA pairs via unsupervised question answering (UQA) methods. However, existing UQA tasks are based on short documents, and can hardly incorporate ...
http://arxiv.org/pdf/2305.02235v1
2023-05-03T16:16:14
2023-05-03T16:16:14
{ "title": [ -0.0107803345, -0.004737854, -0.015022278, -0.045715332, -0.017791748, 0.0077934265, -0.03353882, 0.007949829, 0.018188477, 0.019760132, -0.026672363, 0.0029907227, -0.020309448, 0.0002322197, -0.041503906, -0.0101623535, 0.0025348663, ...
http://arxiv.org/abs/2305.02220v2
WangLab at MEDIQA-Chat 2023: Clinical Note Generation from Doctor-Patient Conversations using Large Language Models
John Giorgi, Augustin Toma, Ronald Xie, Sondra S. Chen, Kevin R. An, Grace X. Zheng, Bo Wang
cs.CL, cs.AI, cs.LG
This paper describes our submission to the MEDIQA-Chat 2023 shared task for automatic clinical note generation from doctor-patient conversations. We report results for two approaches: the first fine-tunes a pre-trained language model (PLM) on the shared task data, and the second uses few-shot in-context learning (ICL) ...
http://arxiv.org/pdf/2305.02220v2
2023-06-03T17:56:29
2023-05-03T15:58:28
{ "title": [ -0.03955078, -0.03186035, -0.02772522, -0.029418945, -0.05557251, 0.0044021606, -0.08630371, -0.003162384, -0.017364502, 0.04650879, -0.0031585693, -0.0021038055, 0.022750854, -0.023208618, -0.02935791, -0.012931824, 0.0051574707, 0....
http://arxiv.org/abs/2305.11873v1
Judgments of research co-created by generative AI: experimental evidence
Paweł Niszczota, Paul Conway
cs.HC, cs.AI, cs.CL, cs.CY, econ.GN, q-fin.EC, K.4.2; I.2.7
The introduction of ChatGPT has fuelled a public debate on the use of generative AI (large language models; LLMs), including its use by researchers. In the current work, we test whether delegating parts of the research process to LLMs leads people to distrust and devalue researchers and scientific output. Participants ...
http://arxiv.org/pdf/2305.11873v1
2023-05-03T15:57:39
2023-05-03T15:57:39
{ "title": [ -0.015411377, -0.0028972626, -0.035003662, 0.0005812645, -0.04220581, -0.03277588, -0.0031852722, -0.033050537, -0.025741577, 0.025482178, -0.0206604, 0.013717651, -0.01890564, -0.033050537, -0.020553589, -0.021987915, -0.011230469, ...
http://arxiv.org/abs/2305.02215v1
Exploring Linguistic Properties of Monolingual BERTs with Typological Classification among Languages
Federico Ranaldi, Elena Sofia Ruzzetti, Felicia Logozzo, Michele Mastromattei, Leonardo Ranaldi, Fabio Massimo Zanzotto
cs.CL, cs.AI
The overwhelming success of transformers is a real conundrum stimulating a compelling question: are these machines replicating some traditional linguistic models or discovering radically new theories? In this paper, we propose a novel standpoint to investigate this important question. Using typological similarities amo...
http://arxiv.org/pdf/2305.02215v1
2023-05-03T15:52:17
2023-05-03T15:52:17
{ "title": [ -0.010803223, -0.004878998, -0.024597168, -0.040649414, -0.015777588, -0.0029468536, -0.0513916, 0.002960205, -0.037902832, 0.048675537, 0.017303467, 0.012901306, 0.018859863, -0.043945312, -0.0034542084, 0.019302368, -0.030227661, 0...
http://arxiv.org/abs/2305.02176v2
Towards Being Parameter-Efficient: A Stratified Sparsely Activated Transformer with Dynamic Capacity
Haoran Xu, Maha Elbayad, Kenton Murray, Jean Maillard, Vedanuj Goswami
cs.CL
Mixture-of-experts (MoE) models that employ sparse activation have demonstrated effectiveness in significantly increasing the number of parameters while maintaining low computational requirements per token. However, recent studies have established that MoE models are inherently parameter-inefficient as the improvement ...
http://arxiv.org/pdf/2305.02176v2
2023-10-22T21:09:23
2023-05-03T15:18:18
{ "title": [ 0.027496338, 0.01890564, -0.035003662, -0.07098389, 0.046081543, -0.010673523, 0.019058228, -0.009338379, -0.004131317, 0.037261963, 0.008453369, 0.009986877, 0.017944336, -0.026565552, -0.020263672, 0.006702423, 0.014694214, 0.03976...
http://arxiv.org/abs/2305.02170v3
A Statistical Exploration of Text Partition Into Constituents: The Case of the Priestly Source in the Books of Genesis and Exodus
Gideon Yoffe, Axel Bühler, Nachum Dershowitz, Israel Finkelstein, Eli Piasetzky, Thomas Römer, Barak Sober
cs.CL
We present a pipeline for a statistical textual exploration, offering a stylometry-based explanation and statistical validation of a hypothesized partition of a text. Given a parameterization of the text, our pipeline: (1) detects literary features yielding the optimal overlap between the hypothesized and unsupervised ...
http://arxiv.org/pdf/2305.02170v3
2023-06-10T07:57:22
2023-05-03T15:07:42
{ "title": [ -0.005706787, -0.012557983, -0.035339355, -0.029418945, -0.008903503, -0.011383057, -0.02859497, -0.0018510818, -0.0075569153, 0.031158447, -0.021942139, 0.042663574, 0.037384033, -0.024032593, 0.004917145, 0.007095337, -0.010643005, ...
http://arxiv.org/abs/2305.02160v1
Explaining Language Models' Predictions with High-Impact Concepts
Ruochen Zhao, Shafiq Joty, Yongjie Wang, Tan Wang
cs.CL
The emergence of large-scale pretrained language models has posed unprecedented challenges in deriving explanations of why the model has made some predictions. Stemmed from the compositional nature of languages, spurious correlations have further undermined the trustworthiness of NLP systems, leading to unreliable mode...
http://arxiv.org/pdf/2305.02160v1
2023-05-03T14:48:27
2023-05-03T14:48:27
{ "title": [ -0.029724121, -0.018844604, -0.020904541, -0.08569336, -0.021331787, 0.018844604, -0.06842041, 0.0057640076, 0.011138916, 0.020706177, 0.0023212433, -0.0060157776, -0.0059013367, -0.021392822, -0.039367676, 0.023712158, -0.0059890747, ...
http://arxiv.org/abs/2305.02156v1
Zero-Shot Listwise Document Reranking with a Large Language Model
Xueguang Ma, Xinyu Zhang, Ronak Pradeep, Jimmy Lin
cs.IR, cs.CL
Supervised ranking methods based on bi-encoder or cross-encoder architectures have shown success in multi-stage text ranking tasks, but they require large amounts of relevance judgments as training data. In this work, we propose Listwise Reranker with a Large Language Model (LRL), which achieves strong reranking effect...
http://arxiv.org/pdf/2305.02156v1
2023-05-03T14:45:34
2023-05-03T14:45:34
{ "title": [ -0.0008125305, -0.0065727234, -0.018966675, -0.046417236, -0.006122589, 0.005935669, -0.035491943, -0.0121536255, 0.017562866, 0.03778076, -0.02519226, -0.00086832047, 0.016647339, -0.0011863708, -0.040130615, -0.0027103424, -0.002168655...
http://arxiv.org/abs/2305.02151v1
Identifying the Correlation Between Language Distance and Cross-Lingual Transfer in a Multilingual Representation Space
Fred Philippy, Siwen Guo, Shohreh Haddadan
cs.CL, cs.AI, cs.LG
Prior research has investigated the impact of various linguistic features on cross-lingual transfer performance. In this study, we investigate the manner in which this effect can be mapped onto the representation space. While past studies have focused on the impact on cross-lingual alignment in multilingual language mo...
http://arxiv.org/pdf/2305.02151v1
2023-05-03T14:33:23
2023-05-03T14:33:23
{ "title": [ -0.0065574646, -0.0007071495, -0.04107666, -0.026947021, 0.008338928, -0.026992798, -0.009757996, -0.027313232, -0.0058631897, 0.03765869, -0.0044898987, 0.025878906, 0.031677246, -0.023986816, -0.016311646, 0.015335083, -0.0035648346, ...
http://arxiv.org/abs/2305.02139v1
A Curriculum View of Robust Loss Functions
Zebin Ou, Yue Zhang
cs.LG, cs.CL
Robust loss functions are designed to combat the adverse impacts of label noise, whose robustness is typically supported by theoretical bounds agnostic to the training dynamics. However, these bounds may fail to characterize the empirical performance as it remains unclear why robust loss functions can underfit. We show...
http://arxiv.org/pdf/2305.02139v1
2023-05-03T14:13:03
2023-05-03T14:13:03
{ "title": [ -0.00687027, 0.00793457, -0.030792236, -0.038116455, 0.010765076, -0.017456055, -0.010604858, -0.028839111, -0.0068511963, 0.0046310425, -0.028335571, 0.014389038, -0.05883789, -0.015235901, -0.019515991, -0.016113281, 0.0033855438, ...
http://arxiv.org/abs/2305.02118v2
Pay More Attention to Relation Exploration for Knowledge Base Question Answering
Yong Cao, Xianzhi Li, Huiwen Liu, Wen Dai, Shuai Chen, Bin Wang, Min Chen, Daniel Hershcovich
cs.CL
Knowledge base question answering (KBQA) is a challenging task that aims to retrieve correct answers from large-scale knowledge bases. Existing attempts primarily focus on entity representation and final answer reasoning, which results in limited supervision for this task. Moreover, the relations, which empirically det...
http://arxiv.org/pdf/2305.02118v2
2023-05-25T10:15:28
2023-05-03T13:48:30
{ "title": [ 0.006385803, -0.00016963482, -0.040527344, -0.06222534, -0.013206482, -0.027770996, -0.008506775, 0.004611969, 0.007873535, 0.013954163, -0.05126953, -0.007675171, 0.00028300285, -0.019851685, -0.012023926, -0.01071167, 0.0067596436, ...
http://arxiv.org/abs/2305.02105v2
GPT-RE: In-context Learning for Relation Extraction using Large Language Models
Zhen Wan, Fei Cheng, Zhuoyuan Mao, Qianying Liu, Haiyue Song, Jiwei Li, Sadao Kurohashi
cs.CL
In spite of the potential for ground-breaking achievements offered by large language models (LLMs) (e.g., GPT-3), they still lag significantly behind fully-supervised baselines (e.g., fine-tuned BERT) in relation extraction (RE). This is due to the two major shortcomings of LLMs in RE: (1) low relevance regarding entit...
http://arxiv.org/pdf/2305.02105v2
2023-10-22T09:13:25
2023-05-03T13:28:08
{ "title": [ -0.0066604614, -0.009643555, -0.04611206, -0.08013916, 0.0093307495, 0.009765625, -0.05279541, 0.003572464, 0.032958984, 0.018356323, -0.030914307, -0.0009899139, 0.008476257, -0.0065956116, -0.043395996, 0.007045746, -0.01727295, 0....
http://arxiv.org/abs/2305.02104v1
Background Knowledge Grounding for Readable, Relevant, and Factual Biomedical Lay Summaries
Domenic Rosati
cs.CL
Communication of scientific findings to the public is important for keeping non-experts informed of developments such as life-saving medical treatments. However, generating readable lay summaries from scientific documents is challenging, and currently, these summaries suffer from critical factual errors. One popular in...
http://arxiv.org/pdf/2305.02104v1
2023-05-03T13:24:49
2023-05-03T13:24:49
{ "title": [ 0.023483276, -0.0068969727, -0.040039062, -0.11383057, -0.022613525, 0.008361816, -0.07476807, 0.028457642, -0.017089844, 0.036102295, -0.010032654, 0.009506226, 0.033843994, -0.022033691, -0.0013561249, 0.0031814575, 0.021957397, 0....
http://arxiv.org/abs/2305.02101v1
What makes a good pause? Investigating the turn-holding effects of fillers
Bing'er Jiang, Erik Ekstedt, Gabriel Skantze
cs.CL
Filled pauses (or fillers), such as "uh" and "um", are frequent in spontaneous speech and can serve as a turn-holding cue for the listener, indicating that the current speaker is not done yet. In this paper, we use the recently proposed Voice Activity Projection (VAP) model, which is a deep learning model trained to pr...
http://arxiv.org/pdf/2305.02101v1
2023-05-03T13:15:37
2023-05-03T13:15:37
{ "title": [ 0.019973755, -0.016296387, -0.01663208, -0.04031372, -0.014823914, -0.033935547, -0.022857666, -0.025253296, -0.0025577545, 0.008407593, -0.059539795, 0.002450943, -0.021972656, -0.014442444, -0.023544312, -0.03842163, -0.008758545, ...
http://arxiv.org/abs/2305.02081v1
Considerations for Ethical Speech Recognition Datasets
Orestis Papakyriakopoulos, Alice Xiang
cs.CY, cs.CL, cs.SD, eess.AS
Speech AI Technologies are largely trained on publicly available datasets or by the massive web-crawling of speech. In both cases, data acquisition focuses on minimizing collection effort, without necessarily taking the data subjects' protection or user needs into consideration. This results to models that are not robu...
http://arxiv.org/pdf/2305.02081v1
2023-05-03T12:38:14
2023-05-03T12:38:14
{ "title": [ 0.013587952, 0.011833191, -0.055267334, -0.05999756, 0.0099487305, -0.037109375, -0.013435364, 0.012649536, -0.033599854, 0.04284668, -0.055908203, 0.0004439354, -0.020584106, -0.008430481, -0.03744507, -0.024871826, -0.013221741, 0....
http://arxiv.org/abs/2305.02036v1
Response-conditioned Turn-taking Prediction
Bing'er Jiang, Erik Ekstedt, Gabriel Skantze
cs.CL, cs.LG
Previous approaches to turn-taking and response generation in conversational systems have treated it as a two-stage process: First, the end of a turn is detected (based on conversation history), then the system generates an appropriate response. Humans, however, do not take the turn just because it is likely, but also ...
http://arxiv.org/pdf/2305.02036v1
2023-05-03T11:06:50
2023-05-03T11:06:50
{ "title": [ 0.054840088, 0.009757996, -0.017547607, -0.045806885, -0.0126571655, -0.028060913, -0.002746582, -0.021408081, -0.015319824, 0.028640747, -0.048797607, 0.0068092346, -0.021087646, -0.015312195, -0.019927979, -0.03186035, 0.008422852, ...
http://arxiv.org/abs/2305.02031v2
A Systematic Study of Knowledge Distillation for Natural Language Generation with Pseudo-Target Training
Nitay Calderon, Subhabrata Mukherjee, Roi Reichart, Amir Kantor
cs.CL, cs.AI
Modern Natural Language Generation (NLG) models come with massive computational and storage requirements. In this work, we study the potential of compressing them, which is crucial for real-world applications serving millions of users. We focus on Knowledge Distillation (KD) techniques, in which a small student model l...
http://arxiv.org/pdf/2305.02031v2
2023-05-26T11:11:11
2023-05-03T10:49:38
{ "title": [ -0.03527832, -0.016571045, -0.06237793, -0.07006836, -0.0065994263, 0.0007529259, -0.06347656, 0.021499634, 0.0061454773, 0.015655518, -0.007217407, -0.003583908, -0.0005559921, -0.029785156, -0.024246216, -0.010025024, -0.011856079, ...
http://arxiv.org/abs/2305.02029v1
Natural language processing on customer note data
Andrew Hilditch, David Webb, Jozef Baca, Tom Armitage, Matthew Shardlow, Peter Appleby
cs.CL
Automatic analysis of customer data for businesses is an area that is of interest to companies. Business to business data is studied rarely in academia due to the sensitive nature of such information. Applying natural language processing can speed up the analysis of prohibitively large sets of data. This paper addresse...
http://arxiv.org/pdf/2305.02029v1
2023-05-03T10:36:56
2023-05-03T10:36:56
{ "title": [ 0.005832672, -0.021102905, -0.01763916, -0.059020996, -0.041992188, -0.0058784485, -0.038146973, 0.014572144, -0.004096985, 0.034088135, -0.07550049, -0.0143585205, 0.014434814, 0.007698059, -0.032226562, -0.03036499, 0.01361084, 0.0...
http://arxiv.org/abs/2305.03502v1
Explore the difficulty of words and its influential attributes based on the Wordle game
Beibei Liu, Yuanfang Zhang, Shiyu Zhang
cs.CL
We adopt the distribution and expectation of guessing times in game Wordle as metrics to predict the difficulty of words and explore their influence factors. In order to predictthe difficulty distribution, we use Monte Carlo to simulate the guessing process of players and then narrow the gap between raw and actual dist...
http://arxiv.org/pdf/2305.03502v1
2023-05-03T09:24:39
2023-05-03T09:24:39
{ "title": [ -0.022354126, -0.013175964, -0.048950195, 0.0134887695, -0.03213501, -0.024398804, 0.025436401, 0.0034046173, -0.039215088, 0.025375366, -0.010559082, 0.00630188, -0.0005555153, -0.03652954, 0.024887085, -0.027053833, -0.0016288757, ...
http://arxiv.org/abs/2305.01965v1
Analysing the Impact of Audio Quality on the Use of Naturalistic Long-Form Recordings for Infant-Directed Speech Research
María Andrea Cruz Blandón, Alejandrina Cristia, Okko Räsänen
cs.CL, cs.SD, eess.AS
Modelling of early language acquisition aims to understand how infants bootstrap their language skills. The modelling encompasses properties of the input data used for training the models, the cognitive hypotheses and their algorithmic implementations being tested, and the evaluation methodologies to compare models to ...
http://arxiv.org/pdf/2305.01965v1
2023-05-03T08:25:37
2023-05-03T08:25:37
{ "title": [ -0.029815674, 0.006088257, -0.06097412, -0.026992798, -0.011749268, -0.021743774, -0.027923584, 0.00036096573, -0.043548584, 0.02470398, -0.028961182, 0.0256958, -0.018478394, -0.0090789795, -0.025314331, -0.034210205, -0.030670166, ...
http://arxiv.org/abs/2305.01957v1
NorQuAD: Norwegian Question Answering Dataset
Sardana Ivanova, Fredrik Aas Andreassen, Matias Jentoft, Sondre Wold, Lilja Øvrelid
cs.CL
In this paper we present NorQuAD: the first Norwegian question answering dataset for machine reading comprehension. The dataset consists of 4,752 manually created question-answer pairs. We here detail the data collection procedure and present statistics of the dataset. We also benchmark several multilingual and Norwegi...
http://arxiv.org/pdf/2305.01957v1
2023-05-03T08:17:07
2023-05-03T08:17:07
{ "title": [ -0.004322052, -0.012641907, -0.027999878, -0.07019043, 0.011993408, 0.00052928925, -0.03543091, 0.001376152, 0.027267456, 0.02078247, -0.041046143, 0.0038318634, 0.017211914, -0.0036697388, -0.035125732, 0.016494751, -0.006626129, 0....
http://arxiv.org/abs/2305.01954v1
SeqAug: Sequential Feature Resampling as a modality agnostic augmentation method
Efthymios Georgiou, Alexandros Potamianos
cs.CL, cs.LG
Data augmentation is a prevalent technique for improving performance in various machine learning applications. We propose SeqAug, a modality-agnostic augmentation method that is tailored towards sequences of extracted features. The core idea of SeqAug is to augment the sequence by resampling from the underlying feature...
http://arxiv.org/pdf/2305.01954v1
2023-05-03T08:11:25
2023-05-03T08:11:25
{ "title": [ -0.009506226, -0.014602661, -0.03857422, -0.048583984, 0.0071792603, -0.0037136078, -0.019851685, -0.029342651, 0.016036987, 0.038269043, -0.012168884, -0.008880615, 0.018035889, -0.048095703, -0.04498291, 0.024108887, 0.0033931732, ...
http://arxiv.org/abs/2305.01951v3
Can LMs Generalize to Future Data? An Empirical Analysis on Text Summarization
Chi Seng Cheang, Hou Pong Chan, Derek F. Wong, Xuebo Liu, Zhaocong Li, Yanming Sun, Shudong Liu, Lidia S. Chao
cs.CL
Recent pre-trained language models (PLMs) achieve promising results in existing abstractive summarization datasets. However, existing summarization benchmarks overlap in time with the standard pre-training corpora and finetuning datasets. Hence, the strong performance of PLMs may rely on the parametric knowledge that i...
http://arxiv.org/pdf/2305.01951v3
2023-11-02T12:07:48
2023-05-03T08:08:07
{ "title": [ -0.0025997162, 0.009033203, -0.02015686, -0.04901123, -0.00598526, -0.008598328, -0.02909851, -0.021606445, -0.013008118, 0.047180176, -0.035247803, 0.039215088, -0.010772705, 0.019958496, -0.064453125, -0.015304565, -0.009399414, 0....
http://arxiv.org/abs/2305.01938v2
Doc2SoarGraph: Discrete Reasoning over Visually-Rich Table-Text Documents with Semantic-Oriented Hierarchical Graphs
Fengbin Zhu, Chao Wang, Fuli Feng, Zifeng Ren, Moxin Li, Tat-Seng Chua
cs.CL, cs.AI
Discrete reasoning over table-text documents (e.g., financial reports) gains increasing attention in recent two years. Existing works mostly simplify this challenge by manually selecting and transforming document pages to structured tables and paragraphs, hindering their practical application. In this work, we explore ...
http://arxiv.org/pdf/2305.01938v2
2023-05-04T10:02:39
2023-05-03T07:30:32
{ "title": [ -0.031555176, -0.012931824, -0.06518555, -0.093933105, 0.004623413, 0.002620697, -0.07348633, 0.021881104, 0.0030593872, 0.031341553, -0.037139893, 0.0048942566, 0.017562866, -0.008148193, -0.042755127, -0.018814087, -0.00969696, 0.0...
http://arxiv.org/abs/2305.01937v1
Can Large Language Models Be an Alternative to Human Evaluations?
Cheng-Han Chiang, Hung-yi Lee
cs.CL, cs.HC
Human evaluation is indispensable and inevitable for assessing the quality of texts generated by machine learning models or written by humans. However, human evaluation is very difficult to reproduce and its quality is notoriously unstable, hindering fair comparisons among different natural language processing (NLP) mo...
http://arxiv.org/pdf/2305.01937v1
2023-05-03T07:28:50
2023-05-03T07:28:50
{ "title": [ -0.017181396, -0.01133728, -0.003255844, 0.0049705505, -0.06628418, -0.019180298, -0.017486572, -0.008934021, -0.009613037, 0.023880005, -0.027862549, -0.04244995, -0.017349243, -0.033996582, -0.018127441, -0.04232788, -0.014717102, ...
http://arxiv.org/abs/2305.01920v1
Generative Meta-Learning for Zero-Shot Relation Triplet Extraction
Wanli Li, Tieyun Qian
cs.CL
The zero-shot relation triplet extraction (ZeroRTE) task aims to extract relation triplets from a piece of text with unseen relation types. The seminal work adopts the pre-trained generative model to generate synthetic samples for new relations. However, current generative models lack the optimization process of model ...
http://arxiv.org/pdf/2305.01920v1
2023-05-03T06:34:39
2023-05-03T06:34:39
{ "title": [ 0.022262573, -0.005908966, -0.03918457, -0.10522461, -0.0001733303, 0.0014047623, -0.06524658, 0.0011768341, 0.06317139, 0.05267334, 0.0021018982, 0.008178711, -0.0087509155, -0.004085541, -0.09051514, 0.027328491, -0.0007638931, 0.0...
http://arxiv.org/abs/2305.01918v3
Improving Contrastive Learning of Sentence Embeddings from AI Feedback
Qinyuan Cheng, Xiaogui Yang, Tianxiang Sun, Linyang Li, Xipeng Qiu
cs.CL, cs.AI
Contrastive learning has become a popular approach in natural language processing, particularly for the learning of sentence embeddings. However, the discrete nature of natural language makes it difficult to ensure the quality of positive and negative sample pairs generated through data augmentation methods. Although s...
http://arxiv.org/pdf/2305.01918v3
2023-05-20T09:39:29
2023-05-03T06:26:13
{ "title": [ -0.010940552, 0.0070495605, -0.0053520203, -0.031951904, -0.0027160645, -0.004512787, -0.016067505, -0.0289917, 0.03729248, 0.017608643, -0.046295166, 0.023147583, -0.00087690353, 0.00023674965, -0.046020508, 0.0037670135, -0.013084412, ...
http://arxiv.org/abs/2305.01914v1
Causal Interventions-based Few-Shot Named Entity Recognition
Zhen Yang, Yongbin Liu, Chunping Ouyang
cs.CL
Few-shot named entity recognition (NER) systems aims at recognizing new classes of entities based on a few labeled samples. A significant challenge in the few-shot regime is prone to overfitting than the tasks with abundant samples. The heavy overfitting in few-shot learning is mainly led by spurious correlation caused...
http://arxiv.org/pdf/2305.01914v1
2023-05-03T06:11:39
2023-05-03T06:11:39
{ "title": [ 0.009536743, -0.009895325, -0.032562256, -0.0814209, 0.005493164, 0.0028247833, -0.04449463, 0.022567749, -0.011039734, 0.0012245178, -0.020599365, 0.025802612, -0.0008854866, -0.021743774, 0.002374649, -0.013458252, -0.0034122467, 0...