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/2303.06662v2
Fuzzy Alignments in Directed Acyclic Graph for Non-Autoregressive Machine Translation
Zhengrui Ma, Chenze Shao, Shangtong Gui, Min Zhang, Yang Feng
cs.CL, I.2.7
Non-autoregressive translation (NAT) reduces the decoding latency but suffers from performance degradation due to the multi-modality problem. Recently, the structure of directed acyclic graph has achieved great success in NAT, which tackles the multi-modality problem by introducing dependency between vertices. However,...
http://arxiv.org/pdf/2303.06662v2
2023-07-17T07:21:50
2023-03-12T13:51:38
{ "title": [ -0.05227661, 0.018203735, -0.061950684, -0.0395813, -0.011314392, -0.006149292, -0.074279785, 0.0014276505, 0.00038409233, 0.06298828, 0.0039024353, 0.0030288696, 0.04171753, -0.022384644, -0.054870605, 0.015808105, 0.0017347336, 0.0...
http://arxiv.org/abs/2303.06623v2
MWE as WSD: Solving Multiword Expression Identification with Word Sense Disambiguation
Joshua Tanner, Jacob Hoffman
cs.CL
Recent approaches to word sense disambiguation (WSD) utilize encodings of the sense gloss (definition), in addition to the input context, to improve performance. In this work we demonstrate that this approach can be adapted for use in multiword expression (MWE) identification by training models which use gloss and cont...
http://arxiv.org/pdf/2303.06623v2
2023-10-19T03:31:53
2023-03-12T09:35:42
{ "title": [ -0.018814087, -0.011917114, -0.025177002, -0.036468506, 0.006412506, -0.008987427, -0.02217102, -0.008773804, 0.022445679, 0.024551392, -0.0072631836, 0.0008339882, 0.017730713, -0.025756836, -0.045135498, 0.010665894, -0.012428284, ...
http://arxiv.org/abs/2303.06585v1
Improving the Intent Classification accuracy in Noisy Environment
Mohamed Nabih Ali, Alessio Brutti, Daniele Falavigna
cs.CL, cs.SD, eess.AS
Intent classification is a fundamental task in the spoken language understanding field that has recently gained the attention of the scientific community, mainly because of the feasibility of approaching it with end-to-end neural models. In this way, avoiding using intermediate steps, i.e. automatic speech recognition,...
http://arxiv.org/pdf/2303.06585v1
2023-03-12T06:11:44
2023-03-12T06:11:44
{ "title": [ -0.004486084, 0.017929077, -0.04928589, -0.027999878, 0.025863647, -0.03060913, 0.0032749176, -0.029830933, 0.034942627, 0.02142334, -0.034057617, 0.002462387, -0.0076141357, -0.01676941, -0.044036865, 0.00021529198, -0.007041931, 0....
http://arxiv.org/abs/2303.06580v1
Towards General Purpose Medical AI: Continual Learning Medical Foundation Model
Huahui Yi, Ziyuan Qin, Qicheng Lao, Wei Xu, Zekun Jiang, Dequan Wang, Shaoting Zhang, Kang Li
cs.CV, cs.CL, cs.LG
Inevitable domain and task discrepancies in real-world scenarios can impair the generalization performance of the pre-trained deep models for medical data. Therefore, we audaciously propose that we should build a general-purpose medical AI system that can be seamlessly adapted to downstream domains/tasks. Since the dom...
http://arxiv.org/pdf/2303.06580v1
2023-03-12T05:27:22
2023-03-12T05:27:22
{ "title": [ -0.015823364, 0.01637268, -0.041900635, -0.04751587, 0.02885437, -0.021697998, -0.006580353, -0.02116394, -0.013572693, 0.033294678, -0.004058838, 0.0037136078, -0.0034637451, -0.03567505, -0.040283203, 0.003929138, 0.0143966675, 0.0...
http://arxiv.org/abs/2303.06574v2
Diffusion Models for Non-autoregressive Text Generation: A Survey
Yifan Li, Kun Zhou, Wayne Xin Zhao, Ji-Rong Wen
cs.CL
Non-autoregressive (NAR) text generation has attracted much attention in the field of natural language processing, which greatly reduces the inference latency but has to sacrifice the generation accuracy. Recently, diffusion models, a class of latent variable generative models, have been introduced into NAR text genera...
http://arxiv.org/pdf/2303.06574v2
2023-05-13T12:42:49
2023-03-12T05:11:09
{ "title": [ -0.02809143, 0.0012559891, -0.04840088, -0.030883789, -0.0015678406, -0.001499176, -0.038513184, -0.0075912476, 0.016967773, 0.040161133, -0.0155181885, 0.024337769, 0.004383087, -0.0020999908, -0.057617188, 0.009132385, -0.033721924, ...
http://arxiv.org/abs/2303.06565v1
Compressed Heterogeneous Graph for Abstractive Multi-Document Summarization
Miao Li, Jianzhong Qi, Jey Han Lau
cs.CL, cs.AI
Multi-document summarization (MDS) aims to generate a summary for a number of related documents. We propose HGSUM, an MDS model that extends an encoder-decoder architecture, to incorporate a heterogeneous graph to represent different semantic units (e.g., words and sentences) of the documents. This contrasts with exist...
http://arxiv.org/pdf/2303.06565v1
2023-03-12T04:23:54
2023-03-12T04:23:54
{ "title": [ -0.032348633, -0.015205383, -0.0446167, -0.08087158, -0.017608643, 0.01966858, -0.07232666, 0.019256592, 0.019012451, 0.04663086, 0.0010738373, 0.013145447, -0.0046310425, -0.016799927, -0.045776367, 0.0028858185, -0.0005965233, 0.00...
http://arxiv.org/abs/2303.13547v1
A comprehensive evaluation of ChatGPT's zero-shot Text-to-SQL capability
Aiwei Liu, Xuming Hu, Lijie Wen, Philip S. Yu
cs.CL, cs.AI, 68T50, I.2.7
This paper presents the first comprehensive analysis of ChatGPT's Text-to-SQL ability. Given the recent emergence of large-scale conversational language model ChatGPT and its impressive capabilities in both conversational abilities and code generation, we sought to evaluate its Text-to-SQL performance. We conducted exp...
http://arxiv.org/pdf/2303.13547v1
2023-03-12T04:22:01
2023-03-12T04:22:01
{ "title": [ 0.01789856, -0.024093628, -0.025756836, -0.050323486, -0.0023975372, -0.009147644, -0.013626099, 0.016235352, 0.017745972, 0.04925537, -0.028030396, 0.0043754578, 0.002986908, -0.016220093, -0.037994385, 0.0026683807, -0.0075416565, ...
http://arxiv.org/abs/2303.06475v1
Transcription free filler word detection with Neural semi-CRFs
Ge Zhu, Yujia Yan, Juan-Pablo Caceres, Zhiyao Duan
eess.AS, cs.CL
Non-linguistic filler words, such as "uh" or "um", are prevalent in spontaneous speech and serve as indicators for expressing hesitation or uncertainty. Previous works for detecting certain non-linguistic filler words are highly dependent on transcriptions from a well-established commercial automatic speech recognition...
http://arxiv.org/pdf/2303.06475v1
2023-03-11T18:17:03
2023-03-11T18:17:03
{ "title": [ -0.01586914, -0.0022239685, -0.025009155, -0.049224854, -0.015403748, -0.016281128, -0.054870605, -0.011459351, 0.013458252, 0.045776367, -0.04119873, 0.008239746, 0.004447937, -0.002216339, -0.061035156, -0.0006465912, -0.023147583, ...
http://arxiv.org/abs/2303.06458v1
ZeroNLG: Aligning and Autoencoding Domains for Zero-Shot Multimodal and Multilingual Natural Language Generation
Bang Yang, Fenglin Liu, Yuexian Zou, Xian Wu, Yaowei Wang, David A. Clifton
cs.CL, cs.AI, cs.CV
Natural Language Generation (NLG) accepts input data in the form of images, videos, or text and generates corresponding natural language text as output. Existing NLG methods mainly adopt a supervised approach and rely heavily on coupled data-to-text pairs. However, for many targeted scenarios and for non-English langua...
http://arxiv.org/pdf/2303.06458v1
2023-03-11T17:14:33
2023-03-11T17:14:33
{ "title": [ -0.029815674, -0.019424438, -0.06390381, -0.06512451, -0.027923584, 0.0066833496, -0.081726074, 0.020401001, -0.000854969, 0.047546387, 0.01727295, 0.011184692, 0.028045654, -0.026992798, -0.019165039, 0.02923584, -0.019363403, 0.034...
http://arxiv.org/abs/2303.06333v2
Parachute: Evaluating Interactive Human-LM Co-writing Systems
Hua Shen, Tongshuang Wu
cs.HC, cs.AI, cs.CL
A surge of advances in language models (LMs) has led to significant interest in using LMs to build co-writing systems, in which humans and LMs interactively contribute to a shared writing artifact. However, there is a lack of studies assessing co-writing systems in interactive settings. We propose a human-centered eval...
http://arxiv.org/pdf/2303.06333v2
2023-03-25T01:40:34
2023-03-11T07:30:52
{ "title": [ -0.05432129, -0.004310608, -0.022460938, -0.009742737, -0.045654297, 0.004425049, -0.06161499, -0.021224976, -0.019363403, 0.03314209, -0.033691406, 0.02178955, -0.014419556, -0.0026111603, -0.037475586, -0.01878357, -0.024383545, 0....
http://arxiv.org/abs/2303.06296v2
Stabilizing Transformer Training by Preventing Attention Entropy Collapse
Shuangfei Zhai, Tatiana Likhomanenko, Etai Littwin, Dan Busbridge, Jason Ramapuram, Yizhe Zhang, Jiatao Gu, Josh Susskind
cs.LG, cs.AI, cs.CL, cs.CV, stat.ML
Training stability is of great importance to Transformers. In this work, we investigate the training dynamics of Transformers by examining the evolution of the attention layers. In particular, we track the attention entropy for each attention head during the course of training, which is a proxy for model sharpness. We ...
http://arxiv.org/pdf/2303.06296v2
2023-07-25T17:42:37
2023-03-11T03:30:47
{ "title": [ 0.03878784, 0.041870117, -0.012832642, -0.02607727, 0.02835083, -0.023986816, 0.019180298, -0.04788208, -0.015022278, 0.0463562, -0.02041626, 0.028564453, 0.0423584, -0.0087509155, -0.026062012, 0.009231567, 0.005332947, 0.018951416,...
http://arxiv.org/abs/2303.06273v3
Consistency Analysis of ChatGPT
Myeongjun Erik Jang, Thomas Lukasiewicz
cs.CL, cs.AI
ChatGPT has gained a huge popularity since its introduction. Its positive aspects have been reported through many media platforms, and some analyses even showed that ChatGPT achieved a decent grade in professional exams, adding extra support to the claim that AI can now assist and even replace humans in industrial fiel...
http://arxiv.org/pdf/2303.06273v3
2023-11-14T00:20:20
2023-03-11T01:19:01
{ "title": [ -0.011070251, 0.002483368, -0.030822754, -0.034301758, -0.0075187683, 0.011375427, -0.033294678, -0.008026123, -0.013427734, 0.029953003, -0.010734558, 0.009864807, 0.006855011, -0.03326416, -0.0034370422, 0.0146865845, -0.017333984, ...
http://arxiv.org/abs/2303.06264v1
An Interactive UI to Support Sensemaking over Collections of Parallel Texts
Joyce Zhou, Elena Glassman, Daniel S. Weld
cs.HC, cs.CL
Scientists and science journalists, among others, often need to make sense of a large number of papers and how they compare with each other in scope, focus, findings, or any other important factors. However, with a large corpus of papers, it's cognitively demanding to pairwise compare and contrast them all with each ot...
http://arxiv.org/pdf/2303.06264v1
2023-03-11T01:04:25
2023-03-11T01:04:25
{ "title": [ -0.004840851, -0.030838013, -0.05456543, -0.031021118, -0.048553467, -0.018798828, -0.03866577, 0.034423828, 0.0026416779, 0.06915283, -0.016616821, 0.0042533875, 0.037139893, -0.011054993, -0.047698975, -0.014678955, 0.018112183, 0....
http://arxiv.org/abs/2303.08080v1
Verbal behavior without syntactic structures: beyond Skinner and Chomsky
Shimon Edelman
cs.CL, q-bio.NC
What does it mean to know language? Since the Chomskian revolution, one popular answer to this question has been: to possess a generative grammar that exclusively licenses certain syntactic structures. Decades later, not even an approximation to such a grammar, for any language, has been formulated; the idea that gramm...
http://arxiv.org/pdf/2303.08080v1
2023-03-11T00:01:21
2023-03-11T00:01:21
{ "title": [ -0.0025596619, 0.0020961761, -0.033203125, -0.06311035, 0.029144287, 0.0007405281, -0.0016870499, -0.016921997, -0.008628845, 0.009765625, -0.0023326874, -0.0013580322, -0.043945312, -0.029037476, -0.039031982, -0.002275467, -0.01184845,...
http://arxiv.org/abs/2303.06245v3
AUTODIAL: Efficient Asynchronous Task-Oriented Dialogue Model
Prajjwal Bhargava, Pooyan Amini, Shahin Shayandeh, Chinnadhurai Sankar
cs.CL
As large dialogue models become commonplace in practice, the problems surrounding high compute requirements for training, inference and larger memory footprint still persists. In this work, we present AUTODIAL, a multi-task dialogue model that addresses the challenges of deploying dialogue model. AUTODIAL utilizes para...
http://arxiv.org/pdf/2303.06245v3
2023-06-09T17:18:03
2023-03-10T23:34:14
{ "title": [ -0.033081055, -0.0064735413, -0.049468994, -0.08758545, 0.022735596, 0.008880615, -0.059387207, 0.010406494, 0.012802124, 0.03692627, -0.00039935112, -0.0072631836, 0.013145447, -0.020217896, -0.049560547, 0.0090408325, -0.012817383, ...
http://arxiv.org/abs/2303.06230v1
Generating Query Focused Summaries without Fine-tuning the Transformer-based Pre-trained Models
Deen Abdullah, Shamanth Nayak, Gandharv Suri, Yllias Chali
cs.CL
Fine-tuning the Natural Language Processing (NLP) models for each new data set requires higher computational time associated with increased carbon footprint and cost. However, fine-tuning helps the pre-trained models adapt to the latest data sets; what if we avoid the fine-tuning steps and attempt to generate summaries...
http://arxiv.org/pdf/2303.06230v1
2023-03-10T22:40:15
2023-03-10T22:40:15
{ "title": [ 0.030792236, 0.0093307495, -0.0178833, -0.07366943, -0.008781433, -0.008026123, -0.049865723, -0.021469116, -0.0046539307, 0.05279541, -0.013069153, 0.016860962, -0.0048065186, -0.005340576, -0.037017822, 0.0057907104, 0.004009247, 0...
http://arxiv.org/abs/2303.06182v2
Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference
Haiyang Huang, Newsha Ardalani, Anna Sun, Liu Ke, Hsien-Hsin S. Lee, Anjali Sridhar, Shruti Bhosale, Carole-Jean Wu, Benjamin Lee
cs.DC, cs.AR, cs.CL, cs.LG
Mixture-of-Experts (MoE) models have gained popularity in achieving state-of-the-art performance in a wide range of tasks in computer vision and natural language processing. They effectively expand the model capacity while incurring a minimal increase in computation cost during training. However, deploying such models ...
http://arxiv.org/pdf/2303.06182v2
2023-06-18T01:33:19
2023-03-10T19:30:15
{ "title": [ -0.027267456, -0.0055618286, -0.057281494, -0.068725586, 0.0019721985, -0.008415222, -0.02078247, -0.0027866364, -0.00970459, 0.020767212, -0.006855011, 0.021438599, -0.0071258545, -0.02583313, -0.01348114, -0.0060310364, 0.004398346, ...
http://arxiv.org/abs/2303.06135v2
Rewarding Chatbots for Real-World Engagement with Millions of Users
Robert Irvine, Douglas Boubert, Vyas Raina, Adian Liusie, Ziyi Zhu, Vineet Mudupalli, Aliaksei Korshuk, Zongyi Liu, Fritz Cremer, Valentin Assassi, Christie-Carol Beauchamp, Xiaoding Lu, Thomas Rialan, William Beauchamp
cs.CL, cs.AI, cs.LG
The emergence of pretrained large language models has led to the deployment of a range of social chatbots for chitchat. Although these chatbots demonstrate language ability and fluency, they are not guaranteed to be engaging and can struggle to retain users. This work investigates the development of social chatbots tha...
http://arxiv.org/pdf/2303.06135v2
2023-03-30T18:28:05
2023-03-10T18:53:52
{ "title": [ 0.068603516, -0.017623901, -0.019058228, -0.040374756, -0.024871826, -0.032287598, 0.014335632, 0.0051612854, -0.048217773, 0.051605225, -0.0209198, 0.011329651, 0.015975952, -0.028045654, 0.008102417, -0.03326416, -0.0052108765, 0.0...
http://arxiv.org/abs/2303.06074v1
Susceptibility to Influence of Large Language Models
Lewis D Griffin, Bennett Kleinberg, Maximilian Mozes, Kimberly T Mai, Maria Vau, Matthew Caldwell, Augustine Marvor-Parker
cs.CL, J.4; I.2.m; I.2.7
Two studies tested the hypothesis that a Large Language Model (LLM) can be used to model psychological change following exposure to influential input. The first study tested a generic mode of influence - the Illusory Truth Effect (ITE) - where earlier exposure to a statement (through, for example, rating its interest) ...
http://arxiv.org/pdf/2303.06074v1
2023-03-10T16:53:30
2023-03-10T16:53:30
{ "title": [ 0.0050697327, 0.0024871826, -0.035125732, -0.06225586, -0.012435913, -0.018936157, -0.04296875, -0.009880066, -0.008392334, 0.030944824, -0.015655518, -0.020233154, 0.0062217712, -0.03604126, -0.026367188, 0.0052337646, -0.011245728, ...
http://arxiv.org/abs/2303.11158v1
ChatGPT as the Transportation Equity Information Source for Scientific Writing
Boniphace Kutela, Shoujia Li, Subasish Das, Jinli Liu
cs.IR, cs.CL, A.1
Transportation equity is an interdisciplinary agenda that requires both transportation and social inputs. Traditionally, transportation equity information are sources from public libraries, conferences, televisions, social media, among other. Artificial intelligence (AI) tools including advanced language models such as...
http://arxiv.org/pdf/2303.11158v1
2023-03-10T16:21:54
2023-03-10T16:21:54
{ "title": [ -0.016921997, -0.01586914, -0.0206604, 0.0023860931, -0.02746582, -0.0088272095, 0.013938904, -0.01737976, -0.026977539, 0.03677368, -0.027877808, 0.024414062, -0.005622864, -0.024505615, -0.0118637085, 0.008132935, -0.014846802, 0.0...
http://arxiv.org/abs/2303.06002v1
Is In-hospital Meta-information Useful for Abstractive Discharge Summary Generation?
Kenichiro Ando, Mamoru Komachi, Takashi Okumura, Hiromasa Horiguchi, Yuji Matsumoto
cs.CL
During the patient's hospitalization, the physician must record daily observations of the patient and summarize them into a brief document called "discharge summary" when the patient is discharged. Automated generation of discharge summary can greatly relieve the physicians' burden, and has been addressed recently in t...
http://arxiv.org/pdf/2303.06002v1
2023-03-10T16:03:19
2023-03-10T16:03:19
{ "title": [ 0.0022277832, -0.008033752, -0.056640625, -0.10559082, -0.02709961, 0.0067825317, -0.08050537, 0.015388489, -0.0031585693, 0.05657959, -0.006664276, 0.029541016, -0.0034370422, -0.02305603, -0.04083252, 0.010307312, 0.012672424, 0.02...
http://arxiv.org/abs/2303.05958v1
Robust Knowledge Distillation from RNN-T Models With Noisy Training Labels Using Full-Sum Loss
Mohammad Zeineldeen, Kartik Audhkhasi, Murali Karthick Baskar, Bhuvana Ramabhadran
cs.CL, cs.SD, eess.AS, stat.ML
This work studies knowledge distillation (KD) and addresses its constraints for recurrent neural network transducer (RNN-T) models. In hard distillation, a teacher model transcribes large amounts of unlabelled speech to train a student model. Soft distillation is another popular KD method that distills the output logit...
http://arxiv.org/pdf/2303.05958v1
2023-03-10T14:46:23
2023-03-10T14:46:23
{ "title": [ -0.021499634, 0.012138367, -0.0519104, -0.072631836, 0.010513306, -0.004196167, -0.060455322, 0.00016784668, 0.02960205, 0.011390686, -0.03805542, -0.010757446, -0.019226074, -0.022537231, -0.043884277, -0.020401001, 0.017593384, 0.0...
http://arxiv.org/abs/2304.02478v1
Exploring AI-Generated Text in Student Writing: How Does AI Help?
David James Woo, Hengky Susanto, Chi Ho Yeung, Kai Guo, April Ka Yeng Fung
cs.CL, cs.AI, cs.CY, J.5; K.3.1
English as foreign language_EFL_students' use of text generated from artificial intelligence_AI_natural language generation_NLG_tools may improve their writing quality. However, it remains unclear to what extent AI-generated text in these students' writing might lead to higher-quality writing. We explored 23 Hong Kong ...
http://arxiv.org/pdf/2304.02478v1
2023-03-10T14:36:47
2023-03-10T14:36:47
{ "title": [ -0.04019165, 0.008377075, -0.03062439, 0.0055351257, -0.01576233, -0.009742737, 0.021957397, -0.018188477, -0.032592773, 0.05316162, -0.04534912, 0.035064697, 0.01725769, 0.007160187, -0.026397705, -0.017150879, -0.022262573, 0.02803...
http://arxiv.org/abs/2303.07304v1
Algorithmic Ghost in the Research Shell: Large Language Models and Academic Knowledge Creation in Management Research
Nigel Williams, Stanislav Ivanov, Dimitrios Buhalis
cs.CL, K.3; K.4; K.6
The paper looks at the role of large language models in academic knowledge creation based on a scoping review (2018 to January 2023) of how researchers have previously used the language model GPT to assist in the performance of academic knowledge creation tasks beyond data analysis. These tasks include writing, editing...
http://arxiv.org/pdf/2303.07304v1
2023-03-10T14:25:29
2023-03-10T14:25:29
{ "title": [ -0.052520752, -0.017593384, -0.04522705, 0.0026054382, -0.012374878, -0.007858276, -0.010917664, -0.019348145, -0.043518066, 0.03353882, -0.018707275, 0.0014028549, -0.025558472, 0.0029144287, -0.0597229, -0.031555176, -0.0065727234, ...
http://arxiv.org/abs/2303.05891v1
Creation and evaluation of timelines for longitudinal user posts
Anthony Hills, Adam Tsakalidis, Federico Nanni, Ioannis Zachos, Maria Liakata
cs.CL
There is increasing interest to work with user generated content in social media, especially textual posts over time. Currently there is no consistent way of segmenting user posts into timelines in a meaningful way that improves the quality and cost of manual annotation. Here we propose a set of methods for segmenting ...
http://arxiv.org/pdf/2303.05891v1
2023-03-10T12:58:34
2023-03-10T12:58:34
{ "title": [ 0.05545044, 0.009140015, -0.0435791, -0.031921387, 0.0047187805, -0.055786133, 0.0034828186, -0.012397766, -0.020584106, 0.037475586, -0.022140503, 0.035736084, -0.011222839, -0.020187378, -0.023864746, 0.010223389, 0.0052986145, 0.0...
http://arxiv.org/abs/2303.08014v1
Does ChatGPT resemble humans in language use?
Zhenguang G. Cai, David A. Haslett, Xufeng Duan, Shuqi Wang, Martin J. Pickering
cs.CL
Large language models (LLMs) and LLM-driven chatbots such as ChatGPT have shown remarkable capacities in comprehending and producing language. However, their internal workings remain a black box in cognitive terms, and it is unclear whether LLMs and chatbots can develop humanlike characteristics in language use. Cognit...
http://arxiv.org/pdf/2303.08014v1
2023-03-10T10:47:59
2023-03-10T10:47:59
{ "title": [ -0.031951904, 0.0048980713, -0.02507019, 0.0473938, -0.039978027, -0.00818634, -0.008728027, -0.03289795, -0.028823853, 0.04260254, -0.008377075, 0.022384644, 0.013435364, -0.013183594, 0.013198853, 0.020187378, -0.021316528, 0.04275...
http://arxiv.org/abs/2303.05834v1
An algebraic approach to translating Japanese
Valentin Boboc
cs.CL, math.CT
We use Lambek's pregroups and the framework of compositional distributional models of language ("DisCoCat") to study translations from Japanese to English as pairs of functors. Adding decorations to pregroups we show how to handle word order changes between languages.
http://arxiv.org/pdf/2303.05834v1
2023-03-10T10:21:20
2023-03-10T10:21:20
{ "title": [ -0.04928589, -0.002538681, -0.0023479462, -0.038726807, -0.019195557, 0.015731812, -0.0602417, -0.015777588, 0.009056091, 0.049194336, 0.019897461, -0.011543274, 0.019485474, -0.0385437, -0.061340332, 0.02381897, 0.013496399, -0.0059...
http://arxiv.org/abs/2303.05759v2
An Overview on Language Models: Recent Developments and Outlook
Chengwei Wei, Yun-Cheng Wang, Bin Wang, C. -C. Jay Kuo
cs.CL
Language modeling studies the probability distributions over strings of texts. It is one of the most fundamental tasks in natural language processing (NLP). It has been widely used in text generation, speech recognition, machine translation, etc. Conventional language models (CLMs) aim to predict the probability of lin...
http://arxiv.org/pdf/2303.05759v2
2023-07-03T05:52:04
2023-03-10T07:55:00
{ "title": [ -0.03479004, -0.013175964, -0.035949707, -0.07659912, 0.0039482117, 0.016082764, -0.07672119, 0.014305115, -0.0015268326, 0.040649414, 0.0067825317, -0.009933472, 0.018539429, -0.03112793, -0.057403564, 0.033843994, -0.016571045, 0.0...
http://arxiv.org/abs/2303.05758v1
MIXPGD: Hybrid Adversarial Training for Speech Recognition Systems
Aminul Huq, Weiyi Zhang, Xiaolin Hu
cs.SD, cs.CL, eess.AS
Automatic speech recognition (ASR) systems based on deep neural networks are weak against adversarial perturbations. We propose mixPGD adversarial training method to improve the robustness of the model for ASR systems. In standard adversarial training, adversarial samples are generated by leveraging supervised or unsup...
http://arxiv.org/pdf/2303.05758v1
2023-03-10T07:52:28
2023-03-10T07:52:28
{ "title": [ -0.011070251, 0.0022010803, -0.045806885, -0.060760498, 0.022964478, -0.0053863525, -0.015670776, 0.009429932, -0.050109863, 0.04324341, -0.010368347, 0.030822754, -0.016311646, -0.015899658, -0.028884888, -0.021575928, -0.022476196, ...
http://arxiv.org/abs/2303.05737v4
Clinical BERTScore: An Improved Measure of Automatic Speech Recognition Performance in Clinical Settings
Joel Shor, Ruyue Agnes Bi, Subhashini Venugopalan, Steven Ibara, Roman Goldenberg, Ehud Rivlin
eess.AS, cs.CL, cs.LG, cs.SD
Automatic Speech Recognition (ASR) in medical contexts has the potential to save time, cut costs, increase report accuracy, and reduce physician burnout. However, the healthcare industry has been slower to adopt this technology, in part due to the importance of avoiding medically-relevant transcription mistakes. In thi...
http://arxiv.org/pdf/2303.05737v4
2023-04-28T21:40:15
2023-03-10T06:46:23
{ "title": [ -0.018814087, -0.011260986, -0.03652954, -0.0635376, 0.0027275085, -0.006126404, -0.046936035, -0.002904892, -0.012939453, 0.036071777, -0.018707275, 0.0073394775, 0.0025615692, -0.02708435, -0.05126953, -0.009254456, -0.019821167, 0...
http://arxiv.org/abs/2303.08016v1
Detection of Abuse in Financial Transaction Descriptions Using Machine Learning
Anna Leontjeva, Genevieve Richards, Kaavya Sriskandaraja, Jessica Perchman, Luiz Pizzato
cs.CL, cs.CY, cs.LG, I.2.7; J.4
Since introducing changes to the New Payments Platform (NPP) to include longer messages as payment descriptions, it has been identified that people are now using it for communication, and in some cases, the system was being used as a targeted form of domestic and family violence. This type of tech-assisted abuse poses ...
http://arxiv.org/pdf/2303.08016v1
2023-03-10T06:10:53
2023-03-10T06:10:53
{ "title": [ 0.0013246536, 0.0036830902, -0.04147339, -0.06124878, 0.018325806, -0.02243042, -0.030136108, -0.006931305, -0.014533997, 0.018127441, -0.03491211, -0.00026631355, 0.017608643, -0.021118164, -0.034454346, -0.03213501, -0.024093628, 0...
http://arxiv.org/abs/2303.05707v1
MuLTI: Efficient Video-and-Language Understanding with MultiWay-Sampler and Multiple Choice Modeling
Jiaqi Xu, Bo Liu, Yunkuo Chen, Mengli Cheng, Xing Shi
cs.CV, cs.CL, cs.MM
Video-and-language understanding has a variety of applications in the industry, such as video question answering, text-video retrieval and multi-label classification. Existing video-and-language understanding methods generally adopt heavy multi-modal encoders and feature fusion modules, which consume large amounts of G...
http://arxiv.org/pdf/2303.05707v1
2023-03-10T05:22:39
2023-03-10T05:22:39
{ "title": [ -0.012710571, -0.021774292, -0.03060913, -0.05923462, -0.0042304993, -0.0061836243, -0.031707764, -0.0045433044, -0.001288414, 0.008972168, -0.039794922, 0.012268066, -0.009437561, -0.02319336, -0.01828003, -0.01727295, -0.013137817, ...
http://arxiv.org/abs/2303.05670v1
Logic Against Bias: Textual Entailment Mitigates Stereotypical Sentence Reasoning
Hongyin Luo, James Glass
cs.CL, cs.AI, cs.CY
Due to their similarity-based learning objectives, pretrained sentence encoders often internalize stereotypical assumptions that reflect the social biases that exist within their training corpora. In this paper, we describe several kinds of stereotypes concerning different communities that are present in popular senten...
http://arxiv.org/pdf/2303.05670v1
2023-03-10T02:52:13
2023-03-10T02:52:13
{ "title": [ 0.010520935, -0.0043029785, -0.039916992, -0.075683594, 0.010253906, -0.018417358, -0.021316528, -0.00001180172, -0.013465881, 0.020553589, -0.010955811, 0.013839722, 0.010894775, -0.011398315, -0.020721436, -0.0011548996, -0.025131226, ...
http://arxiv.org/abs/2303.05581v1
Open World Classification with Adaptive Negative Samples
Ke Bai, Guoyin Wang, Jiwei Li, Sunghyun Park, Sungjin Lee, Puyang Xu, Ricardo Henao, Lawrence Carin
cs.CL, cs.AI
Open world classification is a task in natural language processing with key practical relevance and impact. Since the open or {\em unknown} category data only manifests in the inference phase, finding a model with a suitable decision boundary accommodating for the identification of known classes and discrimination of t...
http://arxiv.org/pdf/2303.05581v1
2023-03-09T21:12:46
2023-03-09T21:12:46
{ "title": [ 0.0066947937, 0.014259338, -0.006324768, -0.030563354, 0.009162903, -0.01335144, 0.0059013367, -0.039001465, 0.010375977, 0.004840851, -0.044799805, 0.011016846, -0.039611816, -0.014297485, -0.038085938, -0.023422241, 0.010551453, 0....
http://arxiv.org/abs/2303.05510v1
Planning with Large Language Models for Code Generation
Shun Zhang, Zhenfang Chen, Yikang Shen, Mingyu Ding, Joshua B. Tenenbaum, Chuang Gan
cs.LG, cs.AI, cs.CL, cs.PL
Existing large language model-based code generation pipelines typically use beam search or sampling algorithms during the decoding process. Although the programs they generate achieve high token-matching-based scores, they often fail to compile or generate incorrect outputs. The main reason is that conventional Transfo...
http://arxiv.org/pdf/2303.05510v1
2023-03-09T18:59:47
2023-03-09T18:59:47
{ "title": [ -0.03869629, -0.028167725, -0.02809143, -0.07373047, -0.01083374, 0.028839111, -0.07470703, 0.016860962, -0.0069999695, 0.016571045, -0.02670288, 0.0045700073, -0.022613525, -0.01525116, -0.011367798, -0.014389038, -0.0126953125, 0.0...
http://arxiv.org/abs/2303.05453v1
Personalisation within bounds: A risk taxonomy and policy framework for the alignment of large language models with personalised feedback
Hannah Rose Kirk, Bertie Vidgen, Paul Röttger, Scott A. Hale
cs.CL, cs.CY
Large language models (LLMs) are used to generate content for a wide range of tasks, and are set to reach a growing audience in coming years due to integration in product interfaces like ChatGPT or search engines like Bing. This intensifies the need to ensure that models are aligned with human preferences and do not pr...
http://arxiv.org/pdf/2303.05453v1
2023-03-09T17:52:07
2023-03-09T17:52:07
{ "title": [ -0.014808655, -0.004753113, -0.040740967, -0.0793457, -0.0076065063, 0.008102417, -0.06750488, -0.0038471222, -0.044769287, 0.035949707, -0.007904053, 0.01727295, -0.011451721, -0.028930664, -0.011077881, -0.012107849, -0.0070114136, ...
http://arxiv.org/abs/2303.08033v1
Large Language Models (GPT) Struggle to Answer Multiple-Choice Questions about Code
Jaromir Savelka, Arav Agarwal, Christopher Bogart, Majd Sakr
cs.CL, cs.AI
We analyzed effectiveness of three generative pre-trained transformer (GPT) models in answering multiple-choice question (MCQ) assessments, often involving short snippets of code, from introductory and intermediate programming courses at the postsecondary level. This emerging technology stirs countless discussions of i...
http://arxiv.org/pdf/2303.08033v1
2023-03-09T16:52:12
2023-03-09T16:52:12
{ "title": [ -0.035247803, -0.030410767, -0.020874023, 0.026824951, -0.0703125, -0.018692017, 0.016433716, 0.0025291443, 0.029632568, 0.015975952, -0.0791626, -0.0075645447, 0.0021800995, -0.023635864, 0.0070610046, -0.023773193, -0.0021266937, 0...
http://arxiv.org/abs/2304.12202v1
ChatGPT may Pass the Bar Exam soon, but has a Long Way to Go for the LexGLUE benchmark
Ilias Chalkidis
cs.CL
Following the hype around OpenAI's ChatGPT conversational agent, the last straw in the recent development of Large Language Models (LLMs) that demonstrate emergent unprecedented zero-shot capabilities, we audit the latest OpenAI's GPT-3.5 model, `gpt-3.5-turbo', the first available ChatGPT model, in the LexGLUE benchma...
http://arxiv.org/pdf/2304.12202v1
2023-03-09T16:42:29
2023-03-09T16:42:29
{ "title": [ -0.016159058, -0.035339355, -0.016677856, -0.010406494, -0.049224854, 0.0028953552, -0.024917603, -0.0026073456, -0.028015137, 0.023742676, -0.042022705, 0.03643799, 0.0046463013, -0.017303467, 0.014282227, -0.025802612, -0.028137207, ...
http://arxiv.org/abs/2303.07203v2
On the Robustness of Text Vectorizers
Rémi Catellier, Samuel Vaiter, Damien Garreau
cs.CL, cs.LG, stat.ML
A fundamental issue in machine learning is the robustness of the model with respect to changes in the input. In natural language processing, models typically contain a first embedding layer, transforming a sequence of tokens into vector representations. While the robustness with respect to changes of continuous inputs ...
http://arxiv.org/pdf/2303.07203v2
2023-06-12T12:55:48
2023-03-09T16:37:37
{ "title": [ -0.0107040405, 0.008682251, -0.006034851, -0.044799805, 0.017227173, 0.0045204163, -0.043670654, -0.044555664, 0.002773285, 0.056518555, -0.02104187, 0.0119018555, 0.0027751923, 0.0035247803, -0.070373535, 0.013618469, -0.0076522827, ...
http://arxiv.org/abs/2303.05349v1
Seeing ChatGPT Through Students' Eyes: An Analysis of TikTok Data
Anna-Carolina Haensch, Sarah Ball, Markus Herklotz, Frauke Kreuter
stat.AP, cs.CL, cs.CY
Advanced large language models like ChatGPT have gained considerable attention recently, including among students. However, while the debate on ChatGPT in academia is making waves, more understanding is needed among lecturers and teachers on how students use and perceive ChatGPT. To address this gap, we analyzed the co...
http://arxiv.org/pdf/2303.05349v1
2023-03-09T15:46:54
2023-03-09T15:46:54
{ "title": [ 0.0041236877, 0.0034923553, -0.016296387, 0.0016107559, 0.0058059692, -0.022064209, 0.017715454, -0.032073975, -0.014671326, 0.021224976, -0.020706177, 0.015136719, -0.02406311, -0.0068969727, -0.036956787, -0.021377563, -0.023269653, ...
http://arxiv.org/abs/2303.11176v1
Types of Approaches, Applications and Challenges in the Development of Sentiment Analysis Systems
Kazem Taghandiki, Elnaz Rezaei Ehsan
cs.CL
Today, the web has become a mandatory platform to express users' opinions, emotions and feelings about various events. Every person using his smartphone can give his opinion about the purchase of a product, the occurrence of an accident, the occurrence of a new disease, etc. in blogs and social networks such as (Twitte...
http://arxiv.org/pdf/2303.11176v1
2023-03-09T15:18:34
2023-03-09T15:18:34
{ "title": [ 0.029144287, -0.0022563934, -0.018920898, -0.024856567, -0.000093877316, -0.025863647, 0.008407593, -0.0026760101, 0.0045814514, 0.049072266, -0.046142578, -0.020187378, 0.0065689087, -0.025466919, -0.035705566, -0.028274536, 0.020401001...
http://arxiv.org/abs/2303.05313v2
Refined Vision-Language Modeling for Fine-grained Multi-modal Pre-training
Lisai Zhang, Qingcai Chen, Zhijian Chen, Yunpeng Han, Zhonghua Li, Zhao Cao
cs.CV, cs.CL
Fine-grained supervision based on object annotations has been widely used for vision and language pre-training (VLP). However, in real-world application scenarios, aligned multi-modal data is usually in the image-caption format, which only provides coarse-grained supervision. It is not only cost-expensive but also comp...
http://arxiv.org/pdf/2303.05313v2
2023-05-06T15:36:10
2023-03-09T15:01:12
{ "title": [ -0.0035190582, -0.0005674362, -0.039794922, -0.07635498, 0.006290436, -0.012863159, -0.055999756, -0.006729126, -0.0027198792, 0.04055786, -0.018829346, 0.008338928, 0.019180298, -0.01864624, -0.036346436, -0.012924194, 0.0062332153, ...
http://arxiv.org/abs/2303.05309v1
MixSpeech: Cross-Modality Self-Learning with Audio-Visual Stream Mixup for Visual Speech Translation and Recognition
Xize Cheng, Linjun Li, Tao Jin, Rongjie Huang, Wang Lin, Zehan Wang, Huangdai Liu, Ye Wang, Aoxiong Yin, Zhou Zhao
cs.CV, cs.CL
Multi-media communications facilitate global interaction among people. However, despite researchers exploring cross-lingual translation techniques such as machine translation and audio speech translation to overcome language barriers, there is still a shortage of cross-lingual studies on visual speech. This lack of res...
http://arxiv.org/pdf/2303.05309v1
2023-03-09T14:58:29
2023-03-09T14:58:29
{ "title": [ -0.020736694, 0.011787415, -0.046905518, -0.0395813, -0.003435135, -0.009475708, -0.025634766, -0.00060749054, -0.004611969, 0.049743652, -0.032104492, 0.04208374, 0.0007047653, -0.016708374, -0.056671143, -0.0063476562, 0.0038890839, ...
http://arxiv.org/abs/2303.05295v1
Dynamic Stashing Quantization for Efficient Transformer Training
Guo Yang, Daniel Lo, Robert Mullins, Yiren Zhao
cs.LG, cs.CL, cs.PF
Large Language Models (LLMs) have demonstrated impressive performance on a range of Natural Language Processing (NLP) tasks. Unfortunately, the immense amount of computations and memory accesses required for LLM training makes them prohibitively expensive in terms of hardware cost, and thus challenging to deploy in use...
http://arxiv.org/pdf/2303.05295v1
2023-03-09T14:44:31
2023-03-09T14:44:31
{ "title": [ -0.005962372, 0.025527954, -0.039642334, -0.039123535, 0.02154541, -0.012580872, -0.0017356873, -0.032928467, -0.027648926, 0.056762695, 0.00032377243, 0.017456055, 0.0098724365, -0.017211914, -0.024993896, 0.011665344, 0.022476196, ...
http://arxiv.org/abs/2303.05221v2
SEAM: An Integrated Activation-Coupled Model of Sentence Processing and Eye Movements in Reading
Maximilian M. Rabe, Dario Paape, Daniela Mertzen, Shravan Vasishth, Ralf Engbert
q-bio.NC, cs.CL
Models of eye-movement control during reading, developed largely within psychology, usually focus on visual, attentional, lexical, and motor processes but neglect post-lexical language processing; by contrast, models of sentence comprehension processes, developed largely within psycholinguistics, generally focus only o...
http://arxiv.org/pdf/2303.05221v2
2023-08-14T08:39:37
2023-03-09T12:50:34
{ "title": [ -0.02053833, -0.0069389343, -0.009300232, 0.0013132095, -0.023712158, -0.0093307495, -0.017242432, -0.03503418, 0.014778137, 0.03805542, -0.0075263977, 0.005065918, 0.017028809, -0.008903503, -0.06677246, 0.006225586, -0.021102905, 0...
http://arxiv.org/abs/2303.05208v1
Geometry of Language
Loe Feijs
cs.CL, 03B65 (Primary) 91F20 (Secundary)
In this article, we present a fresh perspective on language, combining ideas from various sources, but mixed in a new synthesis. As in the minimalist program, the question is whether we can formulate an elegant formalism, a universal grammar or a mechanism which explains significant aspects of the human faculty of lang...
http://arxiv.org/pdf/2303.05208v1
2023-03-09T12:22:28
2023-03-09T12:22:28
{ "title": [ -0.032958984, -0.01184082, -0.014808655, -0.024108887, -0.07116699, -0.026428223, -0.068237305, 0.0009293556, -0.02204895, 0.05480957, -0.011695862, -0.014175415, 0.025878906, -0.036621094, 0.01537323, 0.013587952, 0.031707764, 0.019...
http://arxiv.org/abs/2303.09395v1
Text-to-ECG: 12-Lead Electrocardiogram Synthesis conditioned on Clinical Text Reports
Hyunseung Chung, Jiho Kim, Joon-myoung Kwon, Ki-Hyun Jeon, Min Sung Lee, Edward Choi
cs.CL, cs.LG, eess.SP
Electrocardiogram (ECG) synthesis is the area of research focused on generating realistic synthetic ECG signals for medical use without concerns over annotation costs or clinical data privacy restrictions. Traditional ECG generation models consider a single ECG lead and utilize GAN-based generative models. These models...
http://arxiv.org/pdf/2303.09395v1
2023-03-09T11:58:38
2023-03-09T11:58:38
{ "title": [ -0.008964539, 0.013916016, -0.015159607, -0.020217896, 0.0057525635, -0.00048017502, -0.013580322, -0.038024902, -0.009941101, 0.05331421, -0.026901245, 0.018981934, 0.03616333, -0.033966064, -0.0017032623, 0.005760193, 0.0051498413, ...
http://arxiv.org/abs/2303.08039v1
TQ-Net: Mixed Contrastive Representation Learning For Heterogeneous Test Questions
He Zhu, Xihua Li, Xuemin Zhao, Yunbo Cao, Shan Yu
cs.CL, cs.CV
Recently, more and more people study online for the convenience of access to massive learning materials (e.g. test questions/notes), thus accurately understanding learning materials became a crucial issue, which is essential for many educational applications. Previous studies focus on using language models to represent...
http://arxiv.org/pdf/2303.08039v1
2023-03-09T10:55:48
2023-03-09T10:55:48
{ "title": [ 0.006084442, 0.0045394897, -0.023132324, -0.029785156, 0.0058174133, -0.011833191, 0.0031394958, -0.023544312, 0.029037476, 0.034423828, -0.03656006, 0.014884949, -0.02104187, -0.007637024, -0.05569458, -0.010635376, -0.01222229, 0.0...
http://arxiv.org/abs/2303.05160v1
$π$-augmented pregroups and applications to linguistics
Valentin Boboc
cs.CL, math.LO
We enrich pregroups with a mapping which allows us to locally apply precyclic permutations to designated substrings. We prove a normalisation theorem for such algebraic structures and briefly formalise some known applications of pregroups to the analysis of clitic pronouns in certain natural languages.
http://arxiv.org/pdf/2303.05160v1
2023-03-09T10:33:31
2023-03-09T10:33:31
{ "title": [ -0.004337311, 0.0009198189, -0.0340271, -0.10119629, 0.02128601, 0.0073623657, -0.05718994, 0.01586914, 0.007408142, 0.024642944, -0.027694702, 0.0234375, -0.009841919, -0.0069618225, -0.034576416, 0.01335144, -0.032684326, 0.0234069...
http://arxiv.org/abs/2303.05153v1
Can a Frozen Pretrained Language Model be used for Zero-shot Neural Retrieval on Entity-centric Questions?
Yasuto Hoshi, Daisuke Miyashita, Yasuhiro Morioka, Youyang Ng, Osamu Torii, Jun Deguchi
cs.CL, cs.AI
Neural document retrievers, including dense passage retrieval (DPR), have outperformed classical lexical-matching retrievers, such as BM25, when fine-tuned and tested on specific question-answering datasets. However, it has been shown that the existing dense retrievers do not generalize well not only out of domain but ...
http://arxiv.org/pdf/2303.05153v1
2023-03-09T10:12:18
2023-03-09T10:12:18
{ "title": [ 0.028869629, -0.011894226, -0.00932312, 0.0018911362, -0.035125732, -0.031585693, 0.00982666, -0.0051078796, 0.019958496, 0.050750732, 0.0028972626, -0.0056495667, 0.026306152, -0.026367188, -0.03704834, 0.021057129, 0.010498047, 0.0...
http://arxiv.org/abs/2303.05143v1
ESCL: Equivariant Self-Contrastive Learning for Sentence Representations
Jie Liu, Yixuan Liu, Xue Han, Chao Deng, Junlan Feng
cs.CL, cs.LG
Previous contrastive learning methods for sentence representations often focus on insensitive transformations to produce positive pairs, but neglect the role of sensitive transformations that are harmful to semantic representations. Therefore, we propose an Equivariant Self-Contrastive Learning (ESCL) method to make fu...
http://arxiv.org/pdf/2303.05143v1
2023-03-09T09:52:28
2023-03-09T09:52:28
{ "title": [ 0.003955841, 0.0029335022, -0.03427124, -0.05380249, 0.009208679, -0.014152527, -0.03439331, -0.014312744, 0.009902954, 0.028823853, -0.009239197, 0.025939941, -0.017227173, -0.020965576, -0.037384033, 0.011207581, -0.013893127, 0.06...
http://arxiv.org/abs/2303.05093v1
Improving Video Retrieval by Adaptive Margin
Feng He, Qi Wang, Zhifan Feng, Wenbin Jiang, Yajuan Lv, Yong zhu, Xiao Tan
cs.CV, cs.CL
Video retrieval is becoming increasingly important owing to the rapid emergence of videos on the Internet. The dominant paradigm for video retrieval learns video-text representations by pushing the distance between the similarity of positive pairs and that of negative pairs apart from a fixed margin. However, negative ...
http://arxiv.org/pdf/2303.05093v1
2023-03-09T08:07:38
2023-03-09T08:07:38
{ "title": [ 0.026229858, 0.013595581, -0.032073975, -0.032440186, 0.03074646, -0.022842407, 0.026626587, -0.029846191, 0.008544922, 0.05227661, -0.053527832, -0.01121521, -0.008903503, 0.017364502, -0.07281494, -0.02015686, 0.010536194, 0.009803...
http://arxiv.org/abs/2303.05082v1
Dynamic Multi-View Fusion Mechanism For Chinese Relation Extraction
Jing Yang, Bin Ji, Shasha Li, Jun Ma, Long Peng, Jie Yu
cs.CL
Recently, many studies incorporate external knowledge into character-level feature based models to improve the performance of Chinese relation extraction. However, these methods tend to ignore the internal information of the Chinese character and cannot filter out the noisy information of external knowledge. To address...
http://arxiv.org/pdf/2303.05082v1
2023-03-09T07:35:31
2023-03-09T07:35:31
{ "title": [ -0.0047569275, -0.0018119812, -0.03656006, -0.066345215, 0.008415222, 0.0032310486, -0.022827148, 0.011405945, 0.027191162, 0.026321411, -0.029815674, -0.009750366, -0.004207611, -0.017822266, -0.045928955, -0.007080078, -0.00919342, ...
http://arxiv.org/abs/2303.05080v1
Revisiting the relevance of traditional genres: a network analysis of fiction readers' preferences
Taom Sakal, Stephen Proulx
cs.SI, cs.CL
We investigate how well traditional fiction genres like Fantasy, Thriller, and Literature represent readers' preferences. Using user data from Goodreads we construct a book network where two books are strongly linked if the same people tend to read or enjoy them both. We then partition this network into communities of ...
http://arxiv.org/pdf/2303.05080v1
2023-03-09T07:31:56
2023-03-09T07:31:56
{ "title": [ -0.029663086, 0.0049819946, -0.065979004, -0.0018224716, 0.012023926, -0.04006958, 0.016586304, 0.0055770874, -0.049316406, 0.024246216, 0.03768921, 0.0010728836, -0.017089844, -0.036010742, -0.036315918, -0.0075569153, -0.020004272, ...
http://arxiv.org/abs/2303.05077v2
Learning the Legibility of Visual Text Perturbations
Dev Seth, Rickard Stureborg, Danish Pruthi, Bhuwan Dhingra
cs.CL, cs.CV
Many adversarial attacks in NLP perturb inputs to produce visually similar strings ('ergo' $\rightarrow$ '$\epsilon$rgo') which are legible to humans but degrade model performance. Although preserving legibility is a necessary condition for text perturbation, little work has been done to systematically characterize it;...
http://arxiv.org/pdf/2303.05077v2
2023-03-10T19:54:39
2023-03-09T07:22:07
{ "title": [ -0.008758545, 0.004676819, -0.018371582, -0.019683838, -0.024246216, -0.017410278, -0.007972717, -0.018203735, 0.01121521, 0.020324707, -0.04849243, 0.0072135925, 0.0031337738, -0.014732361, -0.03579712, -0.016723633, 0.0023880005, 0...
http://arxiv.org/abs/2303.05063v4
ICL-D3IE: In-Context Learning with Diverse Demonstrations Updating for Document Information Extraction
Jiabang He, Lei Wang, Yi Hu, Ning Liu, Hui Liu, Xing Xu, Heng Tao Shen
cs.CL
Large language models (LLMs), such as GPT-3 and ChatGPT, have demonstrated remarkable results in various natural language processing (NLP) tasks with in-context learning, which involves inference based on a few demonstration examples. Despite their successes in NLP tasks, no investigation has been conducted to assess t...
http://arxiv.org/pdf/2303.05063v4
2023-08-21T03:57:18
2023-03-09T06:24:50
{ "title": [ -0.021392822, -0.0036449432, -0.054229736, -0.06903076, 0.014549255, -0.008125305, -0.027877808, 0.006225586, 0.01133728, 0.02720642, -0.05419922, 0.021987915, -0.005744934, 0.005554199, -0.031799316, -0.014678955, -0.006137848, 0.04...
http://arxiv.org/abs/2303.05046v1
Unsupervised Language agnostic WER Standardization
Satarupa Guha, Rahul Ambavat, Ankur Gupta, Manish Gupta, Rupeshkumar Mehta
cs.CL
Word error rate (WER) is a standard metric for the evaluation of Automated Speech Recognition (ASR) systems. However, WER fails to provide a fair evaluation of human perceived quality in presence of spelling variations, abbreviations, or compound words arising out of agglutination. Multiple spelling variations might be...
http://arxiv.org/pdf/2303.05046v1
2023-03-09T05:50:54
2023-03-09T05:50:54
{ "title": [ -0.044891357, 0.02305603, -0.099487305, -0.0345459, -0.004989624, -0.030517578, -0.07739258, 0.005832672, -0.010757446, 0.048553467, -0.007637024, 0.0010061264, 0.016189575, -0.040924072, 0.0055236816, 0.02432251, 0.0040397644, 0.058...
http://arxiv.org/abs/2303.05034v1
Multi-Stage Coarse-to-Fine Contrastive Learning for Conversation Intent Induction
Caiyuan Chu, Ya Li, Yifan Liu, Jia-Chen Gu, Quan Liu, Yongxin Ge, Guoping Hu
cs.CL
Intent recognition is critical for task-oriented dialogue systems. However, for emerging domains and new services, it is difficult to accurately identify the key intent of a conversation due to time-consuming data annotation and comparatively poor model transferability. Therefore, the automatic induction of dialogue in...
http://arxiv.org/pdf/2303.05034v1
2023-03-09T04:51:27
2023-03-09T04:51:27
{ "title": [ 0.020202637, 0.005012512, -0.032958984, -0.057281494, 0.0262146, -0.015716553, -0.004787445, 0.0028018951, 0.016845703, 0.01838684, -0.028076172, 0.018203735, -0.019714355, -0.022109985, -0.04055786, -0.0068588257, -0.026168823, 0.05...
http://arxiv.org/abs/2303.07199v1
BeamAttack: Generating High-quality Textual Adversarial Examples through Beam Search and Mixed Semantic Spaces
Hai Zhu, Qingyang Zhao, Yuren Wu
cs.CL, cs.CR
Natural language processing models based on neural networks are vulnerable to adversarial examples. These adversarial examples are imperceptible to human readers but can mislead models to make the wrong predictions. In a black-box setting, attacker can fool the model without knowing model's parameters and architecture....
http://arxiv.org/pdf/2303.07199v1
2023-03-09T03:30:52
2023-03-09T03:30:52
{ "title": [ -0.004901886, -0.014953613, -0.000538826, -0.02432251, -0.011566162, 0.015312195, -0.021499634, -0.039611816, -0.018829346, 0.030029297, 0.0024089813, 0.029754639, -0.023284912, -0.019744873, -0.053649902, -0.0115356445, -0.014976501, ...
http://arxiv.org/abs/2303.04953v1
Let's Get Personal: Personal Questions Improve SocialBot Performance in the Alexa Prize
Kevin K. Bowden, Marilyn Walker
cs.CL, cs.HC
There has been an increased focus on creating conversational open-domain dialogue systems in the spoken dialogue community. Unlike traditional dialogue systems, these conversational systems cannot assume any specific information need or domain restrictions, i.e., the only inherent goal is to converse with the user on a...
http://arxiv.org/pdf/2303.04953v1
2023-03-09T00:10:29
2023-03-09T00:10:29
{ "title": [ 0.00749588, -0.0050621033, -0.012634277, -0.0062408447, -0.00061416626, -0.023895264, 0.022750854, -0.024261475, -0.044036865, 0.028213501, -0.0039100647, 0.018630981, 0.0017576218, -0.029953003, 0.0025749207, -0.0047569275, 0.0063705444...
http://arxiv.org/abs/2303.08006v2
Data-Efficient Learning of Natural Language to Linear Temporal Logic Translators for Robot Task Specification
Jiayi Pan, Glen Chou, Dmitry Berenson
cs.CL, cs.RO
To make robots accessible to a broad audience, it is critical to endow them with the ability to take universal modes of communication, like commands given in natural language, and extract a concrete desired task specification, defined using a formal language like linear temporal logic (LTL). In this paper, we present a...
http://arxiv.org/pdf/2303.08006v2
2023-03-21T03:35:10
2023-03-09T00:09:58
{ "title": [ -0.026306152, -0.024734497, -0.0423584, -0.058166504, -0.0077056885, -0.02029419, -0.039367676, 0.015594482, -0.031188965, 0.022033691, -0.024841309, 0.020141602, -0.02142334, -0.012542725, -0.020217896, -0.023590088, 0.0022659302, 0...
http://arxiv.org/abs/2303.07316v1
FaceChat: An Emotion-Aware Face-to-face Dialogue Framework
Deema Alnuhait, Qingyang Wu, Zhou Yu
cs.CL, cs.AI
While current dialogue systems like ChatGPT have made significant advancements in text-based interactions, they often overlook the potential of other modalities in enhancing the overall user experience. We present FaceChat, a web-based dialogue framework that enables emotionally-sensitive and face-to-face conversations...
http://arxiv.org/pdf/2303.07316v1
2023-03-08T20:45:37
2023-03-08T20:45:37
{ "title": [ -0.003293991, -0.0049591064, -0.035247803, -0.055847168, 0.010017395, -0.00017404556, -0.035461426, -0.010169983, 0.037750244, 0.027374268, -0.019348145, -0.006454468, 0.015281677, -0.029541016, -0.046813965, 0.0069274902, -0.009864807, ...
http://arxiv.org/abs/2303.12737v1
Comparing Trajectory and Vision Modalities for Verb Representation
Dylan Ebert, Chen Sun, Ellie Pavlick
cs.CV, cs.AI, cs.CL, 68T50
Three-dimensional trajectories, or the 3D position and rotation of objects over time, have been shown to encode key aspects of verb semantics (e.g., the meanings of roll vs. slide). However, most multimodal models in NLP use 2D images as representations of the world. Given the importance of 3D space in formal models of...
http://arxiv.org/pdf/2303.12737v1
2023-03-08T20:32:42
2023-03-08T20:32:42
{ "title": [ -0.0014467239, 0.0016880035, -0.026107788, -0.030944824, -0.0138168335, -0.029220581, -0.024398804, -0.021972656, 0.014434814, 0.044403076, -0.0385437, -0.00466156, 0.027069092, -0.015609741, -0.037750244, 0.0008034706, 0.0105896, 0....
http://arxiv.org/abs/2303.04851v1
Lexical Complexity Prediction: An Overview
Kai North, Marcos Zampieri, Matthew Shardlow
cs.CL, A.1
The occurrence of unknown words in texts significantly hinders reading comprehension. To improve accessibility for specific target populations, computational modelling has been applied to identify complex words in texts and substitute them for simpler alternatives. In this paper, we present an overview of computational...
http://arxiv.org/pdf/2303.04851v1
2023-03-08T19:35:08
2023-03-08T19:35:08
{ "title": [ -0.026885986, 0.00042796135, -0.03189087, -0.0690918, 0.00056648254, 0.011680603, -0.059539795, 0.004322052, -0.01020813, 0.050109863, -0.00015568733, 0.0018558502, 0.0062561035, -0.01676941, -0.0690918, 0.010559082, -0.017211914, 0....
http://arxiv.org/abs/2303.04838v1
The Casual Conversations v2 Dataset
Bilal Porgali, Vítor Albiero, Jordan Ryda, Cristian Canton Ferrer, Caner Hazirbas
cs.CV, cs.AI, cs.CL, cs.CY
This paper introduces a new large consent-driven dataset aimed at assisting in the evaluation of algorithmic bias and robustness of computer vision and audio speech models in regards to 11 attributes that are self-provided or labeled by trained annotators. The dataset includes 26,467 videos of 5,567 unique paid partici...
http://arxiv.org/pdf/2303.04838v1
2023-03-08T19:17:05
2023-03-08T19:17:05
{ "title": [ -0.013008118, 0.0016860962, -0.036834717, -0.04107666, -0.018463135, -0.020492554, -0.06286621, -0.009597778, -0.004009247, 0.051940918, 0.02041626, -0.016906738, 0.024673462, -0.048187256, 0.0056648254, 0.016616821, 0.010406494, 0.0...
http://arxiv.org/abs/2303.05431v1
disco: a toolkit for Distributional Control of Generative Models
Germán Kruszewski, Jos Rozen, Marc Dymetman
cs.CL, cs.AI, cs.LG
Pre-trained language models and other generative models have revolutionized NLP and beyond. However, these models tend to reproduce undesirable biases present in their training data. Also, they may overlook patterns that are important but challenging to capture. To address these limitations, researchers have introduced...
http://arxiv.org/pdf/2303.05431v1
2023-03-08T18:58:52
2023-03-08T18:58:52
{ "title": [ -0.011054993, -0.011512756, -0.03062439, -0.046173096, 0.010292053, -0.00039720535, -0.031677246, -0.03265381, 0.017959595, 0.017623901, 0.00081014633, 0.015945435, -0.011428833, -0.02911377, -0.036865234, 0.021621704, -0.009986877, ...
http://arxiv.org/abs/2303.04794v1
Comprehensive Event Representations using Event Knowledge Graphs and Natural Language Processing
Tin Kuculo
cs.CL
Recent work has utilised knowledge-aware approaches to natural language understanding, question answering, recommendation systems, and other tasks. These approaches rely on well-constructed and large-scale knowledge graphs that can be useful for many downstream applications and empower knowledge-aware models with commo...
http://arxiv.org/pdf/2303.04794v1
2023-03-08T18:43:39
2023-03-08T18:43:39
{ "title": [ 0.006511688, -0.010147095, -0.03677368, -0.11755371, 0.024047852, 0.016418457, -0.08270264, -0.010612488, 0.014511108, 0.03314209, 0.005077362, -0.011604309, 0.006881714, -0.031799316, -0.037353516, 0.017303467, 0.0025596619, 0.02304...
http://arxiv.org/abs/2303.04729v3
On the Risks of Stealing the Decoding Algorithms of Language Models
Ali Naseh, Kalpesh Krishna, Mohit Iyyer, Amir Houmansadr
cs.LG, cs.CL, cs.CR
A key component of generating text from modern language models (LM) is the selection and tuning of decoding algorithms. These algorithms determine how to generate text from the internal probability distribution generated by the LM. The process of choosing a decoding algorithm and tuning its hyperparameters takes signif...
http://arxiv.org/pdf/2303.04729v3
2023-04-26T03:16:43
2023-03-08T17:15:58
{ "title": [ -0.015701294, -0.013290405, -0.010513306, 0.0008621216, -0.046661377, -0.018157959, -0.015266418, -0.02243042, -0.035980225, 0.05279541, -0.02168274, -0.0013065338, 0.0184021, -0.03237915, -0.014717102, 0.00008028746, -0.0027580261, ...
http://arxiv.org/abs/2303.04715v2
Extending the Pre-Training of BLOOM for Improved Support of Traditional Chinese: Models, Methods and Results
Philipp Ennen, Po-Chun Hsu, Chan-Jan Hsu, Chang-Le Liu, Yen-Chen Wu, Yin-Hsiang Liao, Chin-Tung Lin, Da-Shan Shiu, Wei-Yun Ma
cs.CL, cs.AI
In this paper we present the multilingual language model BLOOM-zh that features enhanced support for Traditional Chinese. BLOOM-zh has its origins in the open-source BLOOM models presented by BigScience in 2022. Starting from released models, we extended the pre-training of BLOOM by additional 7.4 billion tokens in Tra...
http://arxiv.org/pdf/2303.04715v2
2023-06-23T14:54:03
2023-03-08T16:53:19
{ "title": [ 0.012924194, -0.0016222, -0.031280518, -0.091308594, 0.008842468, 0.011657715, -0.05392456, 0.008224487, -0.009712219, 0.043884277, -0.010009766, -0.0033016205, 0.06561279, -0.039367676, -0.025421143, 0.020385742, 0.002904892, 0.0128...
http://arxiv.org/abs/2303.04691v2
Self-contained Beta-with-Spikes Approximation for Inference Under a Wright-Fisher Model
Juan Guerrero Montero, Richard A. Blythe
q-bio.PE, cond-mat.stat-mech, cs.CL
We construct a reliable estimation of evolutionary parameters within the Wright-Fisher model, which describes changes in allele frequencies due to selection and genetic drift, from time-series data. Such data exists for biological populations, for example via artificial evolution experiments, and for the cultural evolu...
http://arxiv.org/pdf/2303.04691v2
2023-05-11T15:59:00
2023-03-08T16:32:10
{ "title": [ 0.0033664703, 0.01398468, -0.0074691772, -0.043518066, 0.010643005, 0.008460999, 0.0003774166, -0.02027893, -0.031555176, 0.011161804, -0.01109314, 0.020629883, -0.021499634, -0.02999878, -0.020584106, -0.016479492, 0.016143799, 0.01...
http://arxiv.org/abs/2303.04673v2
Cost-Effective Hyperparameter Optimization for Large Language Model Generation Inference
Chi Wang, Susan Xueqing Liu, Ahmed H. Awadallah
cs.CL, cs.AI, cs.LG
Large Language Models (LLMs) have sparked significant interest in their generative capabilities, leading to the development of various commercial applications. The high cost of using the models drives application builders to maximize the value of generation under a limited inference budget. This paper presents a study ...
http://arxiv.org/pdf/2303.04673v2
2023-08-08T18:04:11
2023-03-08T15:52:14
{ "title": [ -0.03881836, -0.02368164, -0.036102295, -0.085632324, -0.01322937, 0.022064209, -0.10760498, 0.007610321, -0.006072998, 0.00881958, -0.008880615, 0.006778717, 0.0056533813, -0.033477783, -0.007904053, 0.011871338, -0.0071907043, 0.04...
http://arxiv.org/abs/2303.09461v2
ChatGPT Participates in a Computer Science Exam
Sebastian Bordt, Ulrike von Luxburg
cs.CL, cs.CY
We asked ChatGPT to participate in an undergraduate computer science exam on ''Algorithms and Data Structures''. The program was evaluated on the entire exam as posed to the students. We hand-copied its answers onto an exam sheet, which was subsequently graded in a blind setup alongside those of 200 participating stude...
http://arxiv.org/pdf/2303.09461v2
2023-03-22T11:30:41
2023-03-08T15:46:14
{ "title": [ -0.0005950928, -0.03201294, 0.026535034, 0.003917694, -0.07525635, 0.007583618, -0.00033068657, 0.013519287, -0.019363403, 0.02722168, -0.020202637, 0.0077209473, 0.009666443, -0.029174805, 0.040039062, 0.005340576, 0.02067566, 0.013...
http://arxiv.org/abs/2303.04562v3
Extrapolative Controlled Sequence Generation via Iterative Refinement
Vishakh Padmakumar, Richard Yuanzhe Pang, He He, Ankur P. Parikh
cs.LG, cs.CL, q-bio.QM
We study the problem of extrapolative controlled generation, i.e., generating sequences with attribute values beyond the range seen in training. This task is of significant importance in automated design, especially drug discovery, where the goal is to design novel proteins that are \textit{better} (e.g., more stable) ...
http://arxiv.org/pdf/2303.04562v3
2023-06-07T15:34:38
2023-03-08T13:21:27
{ "title": [ -0.03918457, -0.011924744, -0.0463562, -0.062561035, -0.021194458, 0.005832672, -0.088256836, -0.0067214966, -0.014152527, 0.04547119, -0.036956787, 0.024169922, 0.0048675537, -0.005458832, -0.022628784, -0.008361816, -0.021453857, 0...
http://arxiv.org/abs/2303.04544v1
Models of symbol emergence in communication: a conceptual review and a guide for avoiding local minima
Julian Zubek, Tomasz Korbak, Joanna Rączaszek-Leonardi
cs.AI, cs.CL, cs.MA
Computational simulations are a popular method for testing hypotheses about the emergence of communication. This kind of research is performed in a variety of traditions including language evolution, developmental psychology, cognitive science, machine learning, robotics, etc. The motivations for the models are differe...
http://arxiv.org/pdf/2303.04544v1
2023-03-08T12:53:03
2023-03-08T12:53:03
{ "title": [ -0.010986328, 0.014175415, -0.025970459, 0.0070228577, 0.011741638, -0.005241394, 0.011146545, -0.037719727, -0.021636963, 0.045288086, 0.02218628, 0.006401062, 0.0011205673, -0.032806396, -0.04135132, 0.012931824, -0.0050621033, 0.0...
http://arxiv.org/abs/2303.05387v1
Automatic Detection of Industry Sectors in Legal Articles Using Machine Learning Approaches
Hui Yang, Stella Hadjiantoni, Yunfei Long, Ruta Petraityte, Berthold Lausen
cs.CL, cs.LG, I.2.6, I.2.7, 68T50, 68T07
The ability to automatically identify industry sector coverage in articles on legal developments, or any kind of news articles for that matter, can bring plentiful of benefits both to the readers and the content creators themselves. By having articles tagged based on industry coverage, readers from all around the world...
http://arxiv.org/pdf/2303.05387v1
2023-03-08T12:41:56
2023-03-08T12:41:56
{ "title": [ -0.050994873, 0.017562866, -0.07598877, -0.010787964, 0.006298065, -0.027618408, -0.02658081, -0.021865845, -0.01864624, 0.0062332153, -0.047058105, 0.000385046, -0.03277588, -0.021530151, -0.022369385, -0.032043457, -0.029724121, 0....
http://arxiv.org/abs/2303.04526v2
Student's t-Distribution: On Measuring the Inter-Rater Reliability When the Observations are Scarce
Serge Gladkoff, Lifeng Han, Goran Nenadic
cs.CL, cs.IT, cs.NA, math.IT, math.NA, stat.AP
In natural language processing (NLP) we always rely on human judgement as the golden quality evaluation method. However, there has been an ongoing debate on how to better evaluate inter-rater reliability (IRR) levels for certain evaluation tasks, such as translation quality evaluation (TQE), especially when the data sa...
http://arxiv.org/pdf/2303.04526v2
2023-07-09T16:13:25
2023-03-08T11:51:26
{ "title": [ -0.024902344, 0.032104492, -0.036224365, -0.007171631, 0.023391724, -0.01701355, 0.012870789, -0.05657959, 0.0010938644, 0.0076026917, -0.052947998, -0.00023126602, -0.04055786, -0.01210022, -0.047943115, -0.029159546, -0.014587402, ...
http://arxiv.org/abs/2303.04496v2
MenuCraft: Interactive Menu System Design with Large Language Models
Amir Hossein Kargaran, Nafiseh Nikeghbal, Abbas Heydarnoori, Hinrich Schütze
cs.CL, cs.AI, cs.HC
Menu system design is a challenging task involving many design options and various human factors. For example, one crucial factor that designers need to consider is the semantic and systematic relation of menu commands. However, capturing these relations can be challenging due to limited available resources. With the a...
http://arxiv.org/pdf/2303.04496v2
2023-07-23T21:36:14
2023-03-08T10:39:38
{ "title": [ -0.017089844, -0.008049011, -0.01637268, -0.028686523, 0.019348145, 0.005470276, -0.01878357, -0.030136108, 0.006969452, 0.036132812, -0.01525116, -0.010543823, -0.00548172, -0.012336731, -0.058746338, -0.006088257, -0.005214691, 0.0...
http://arxiv.org/abs/2303.04487v3
Query-Utterance Attention with Joint modeling for Query-Focused Meeting Summarization
Xingxian Liu, Bin Duan, Bo Xiao, Yajing Xu
cs.CL, cs.AI
Query-focused meeting summarization (QFMS) aims to generate summaries from meeting transcripts in response to a given query. Previous works typically concatenate the query with meeting transcripts and implicitly model the query relevance only at the token level with attention mechanism. However, due to the dilution of ...
http://arxiv.org/pdf/2303.04487v3
2023-06-01T03:07:31
2023-03-08T10:21:45
{ "title": [ 0.023651123, 0.023345947, -0.04220581, -0.103149414, 0.014961243, 0.0033435822, -0.06945801, 0.0009946823, 0.023834229, 0.06939697, -0.014350891, 0.009170532, 0.05847168, 0.008224487, -0.06518555, 0.024719238, -0.005355835, -0.003952...
http://arxiv.org/abs/2303.04426v2
NASTyLinker: NIL-Aware Scalable Transformer-based Entity Linker
Nicolas Heist, Heiko Paulheim
cs.CL, cs.AI, cs.IR
Entity Linking (EL) is the task of detecting mentions of entities in text and disambiguating them to a reference knowledge base. Most prevalent EL approaches assume that the reference knowledge base is complete. In practice, however, it is necessary to deal with the case of linking to an entity that is not contained in...
http://arxiv.org/pdf/2303.04426v2
2023-03-13T08:43:27
2023-03-08T08:08:57
{ "title": [ -0.005420685, 0.015701294, -0.064697266, -0.099121094, 0.046661377, 0.0025749207, -0.033172607, 0.008781433, -0.0056648254, 0.039642334, -0.001783371, -0.0073928833, 0.046844482, -0.029129028, -0.008651733, 0.01701355, -0.007896423, ...
http://arxiv.org/abs/2303.04381v1
Automatically Auditing Large Language Models via Discrete Optimization
Erik Jones, Anca Dragan, Aditi Raghunathan, Jacob Steinhardt
cs.LG, cs.CL
Auditing large language models for unexpected behaviors is critical to preempt catastrophic deployments, yet remains challenging. In this work, we cast auditing as an optimization problem, where we automatically search for input-output pairs that match a desired target behavior. For example, we might aim to find a non-...
http://arxiv.org/pdf/2303.04381v1
2023-03-08T05:09:59
2023-03-08T05:09:59
{ "title": [ -0.062042236, -0.021575928, -0.047668457, -0.06640625, -0.02420044, 0.01927185, -0.08251953, 0.011940002, -0.011764526, 0.032684326, -0.024536133, -0.000032901764, -0.0056037903, -0.0064201355, -0.042907715, -0.0158844, -0.015022278, ...
http://arxiv.org/abs/2303.04361v1
Sample Efficient Multimodal Semantic Augmentation for Incremental Summarization
Sumanta Bhattacharyya, Ramesh Manuvinakurike, Sahisnu Mazumder, Saurav Sahay
cs.CL, cs.CV
In this work, we develop a prompting approach for incremental summarization of task videos. We develop a sample-efficient few-shot approach for extracting semantic concepts as an intermediate step. We leverage an existing model for extracting the concepts from the images and extend it to videos and introduce a clusteri...
http://arxiv.org/pdf/2303.04361v1
2023-03-08T03:58:06
2023-03-08T03:58:06
{ "title": [ 0.004573822, 0.009643555, -0.05822754, -0.095825195, 0.00793457, -0.0060043335, -0.061035156, 0.014846802, 0.026916504, 0.041137695, -0.026275635, 0.027328491, 0.011314392, 0.007926941, -0.038146973, 0.013244629, -0.0037269592, 0.024...
http://arxiv.org/abs/2303.04360v2
Does Synthetic Data Generation of LLMs Help Clinical Text Mining?
Ruixiang Tang, Xiaotian Han, Xiaoqian Jiang, Xia Hu
cs.CL, cs.AI, cs.LG
Recent advancements in large language models (LLMs) have led to the development of highly potent models like OpenAI's ChatGPT. These models have exhibited exceptional performance in a variety of tasks, such as question answering, essay composition, and code generation. However, their effectiveness in the healthcare sec...
http://arxiv.org/pdf/2303.04360v2
2023-04-10T18:47:51
2023-03-08T03:56:31
{ "title": [ -0.060577393, -0.013053894, -0.044311523, -0.027496338, -0.047027588, 0.01272583, -0.0713501, 0.008399963, -0.034179688, 0.013870239, -0.051116943, 0.034576416, -0.041259766, -0.013084412, -0.0017881393, -0.042266846, -0.022903442, 0...
http://arxiv.org/abs/2303.04289v1
Do Prosody Transfer Models Transfer Prosody?
Atli Thor Sigurgeirsson, Simon King
cs.CL, cs.SD, eess.AS
Some recent models for Text-to-Speech synthesis aim to transfer the prosody of a reference utterance to the generated target synthetic speech. This is done by using a learned embedding of the reference utterance, which is used to condition speech generation. During training, the reference utterance is identical to the ...
http://arxiv.org/pdf/2303.04289v1
2023-03-07T23:35:58
2023-03-07T23:35:58
{ "title": [ -0.014381409, 0.0067749023, -0.02911377, -0.032806396, -0.03503418, -0.001543045, -0.040985107, 0.0060043335, -0.022506714, 0.062683105, 0.013824463, -0.01676941, 0.046661377, -0.046081543, -0.020645142, 0.005760193, -0.011985779, 0....
http://arxiv.org/abs/2303.05279v1
Can large language models build causal graphs?
Stephanie Long, Tibor Schuster, Alexandre Piché, Department of Family Medicine, McGill University, Mila, Université de Montreal, ServiceNow Research
cs.CL, cs.AI
Building causal graphs can be a laborious process. To ensure all relevant causal pathways have been captured, researchers often have to discuss with clinicians and experts while also reviewing extensive relevant medical literature. By encoding common and medical knowledge, large language models (LLMs) represent an oppo...
http://arxiv.org/pdf/2303.05279v1
2023-03-07T22:05:31
2023-03-07T22:05:31
{ "title": [ -0.030380249, -0.020004272, -0.03982544, -0.0960083, -0.03390503, 0.01676941, -0.10241699, 0.031173706, -0.02079773, 0.001449585, -0.00894928, -0.0029888153, -0.0181427, -0.034851074, 0.005844116, -0.0030403137, -0.010658264, 0.06015...
http://arxiv.org/abs/2303.04245v2
How Do Transformers Learn Topic Structure: Towards a Mechanistic Understanding
Yuchen Li, Yuanzhi Li, Andrej Risteski
cs.LG, cs.CL, stat.ML
While the successes of transformers across many domains are indisputable, accurate understanding of the learning mechanics is still largely lacking. Their capabilities have been probed on benchmarks which include a variety of structured and reasoning tasks -- but mathematical understanding is lagging substantially behi...
http://arxiv.org/pdf/2303.04245v2
2023-07-24T17:29:04
2023-03-07T21:42:17
{ "title": [ -0.020248413, 0.008995056, -0.032287598, -0.026000977, 0.0057792664, -0.005832672, -0.021194458, -0.025009155, -0.012023926, 0.014961243, -0.036834717, 0.037200928, -0.023162842, -0.022277832, -0.007583618, -0.0025100708, -0.0076789856, ...
http://arxiv.org/abs/2303.04226v1
A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT
Yihan Cao, Siyu Li, Yixin Liu, Zhiling Yan, Yutong Dai, Philip S. Yu, Lichao Sun
cs.AI, cs.CL, cs.LG
Recently, ChatGPT, along with DALL-E-2 and Codex,has been gaining significant attention from society. As a result, many individuals have become interested in related resources and are seeking to uncover the background and secrets behind its impressive performance. In fact, ChatGPT and other Generative AI (GAI) techniqu...
http://arxiv.org/pdf/2303.04226v1
2023-03-07T20:36:13
2023-03-07T20:36:13
{ "title": [ -0.018722534, -0.00945282, -0.032409668, -0.034118652, -0.004043579, -0.012992859, -0.00856781, -0.01940918, 0.00504303, 0.030227661, -0.027618408, 0.024490356, -0.008972168, -0.017211914, -0.032043457, 0.012313843, -0.012748718, 0.0...
http://arxiv.org/abs/2303.04222v2
SemEval-2023 Task 10: Explainable Detection of Online Sexism
Hannah Rose Kirk, Wenjie Yin, Bertie Vidgen, Paul Röttger
cs.CL, cs.CY
Online sexism is a widespread and harmful phenomenon. Automated tools can assist the detection of sexism at scale. Binary detection, however, disregards the diversity of sexist content, and fails to provide clear explanations for why something is sexist. To address this issue, we introduce SemEval Task 10 on the Explai...
http://arxiv.org/pdf/2303.04222v2
2023-05-08T14:34:49
2023-03-07T20:28:39
{ "title": [ 0.008041382, -0.00687027, -0.03942871, -0.030227661, 0.015487671, -0.036346436, 0.013244629, -0.011413574, 0.0022583008, 0.039794922, -0.0060272217, 0.010543823, 0.00944519, -0.014701843, -0.03756714, 0.0013093948, -0.02748108, 0.049...
http://arxiv.org/abs/2303.04185v2
Gradient-Free Structured Pruning with Unlabeled Data
Azade Nova, Hanjun Dai, Dale Schuurmans
cs.LG, cs.AI, cs.CL
Large Language Models (LLMs) have achieved great success in solving difficult tasks across many domains, but such success comes with a high computation cost, and inference latency. As developers and third parties customize these models, the need to provide efficient inference has increased. Many efforts have attempted ...
http://arxiv.org/pdf/2303.04185v2
2023-07-15T20:19:22
2023-03-07T19:12:31
{ "title": [ -0.01586914, 0.014503479, -0.038909912, -0.08795166, 0.010917664, 0.02507019, -0.05960083, -0.0011386871, 0.029754639, -0.0049934387, -0.030334473, 0.014144897, -0.0418396, -0.019470215, -0.008773804, -0.0068130493, 0.01424408, 0.023...
http://arxiv.org/abs/2303.04134v2
A Hybrid Architecture for Out of Domain Intent Detection and Intent Discovery
Masoud Akbari, Ali Mohades, M. Hassan Shirali-Shahreza
cs.CL, cs.AI
Intent Detection is one of the tasks of the Natural Language Understanding (NLU) unit in task-oriented dialogue systems. Out of Scope (OOS) and Out of Domain (OOD) inputs may run these systems into a problem. On the other side, a labeled dataset is needed to train a model for Intent Detection in task-oriented dialogue ...
http://arxiv.org/pdf/2303.04134v2
2023-07-30T16:38:23
2023-03-07T18:49:13
{ "title": [ -0.000603199, 0.017654419, -0.0368042, -0.03479004, 0.04623413, -0.02519226, 0.0070648193, -0.02949524, -0.007068634, 0.018081665, -0.026687622, 0.011436462, -0.003271103, -0.016601562, -0.029846191, -0.009101868, -0.03237915, 0.0480...
http://arxiv.org/abs/2303.04132v2
Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction
Martin Josifoski, Marija Sakota, Maxime Peyrard, Robert West
cs.CL, cs.AI, cs.LG
Large language models (LLMs) have great potential for synthetic data generation. This work shows that useful data can be synthetically generated even for tasks that cannot be solved directly by LLMs: for problems with structured outputs, it is possible to prompt an LLM to perform the task in the reverse direction, by g...
http://arxiv.org/pdf/2303.04132v2
2023-10-29T14:24:46
2023-03-07T18:48:55
{ "title": [ -0.050750732, -0.010543823, -0.053344727, -0.018493652, -0.026550293, -0.014091492, -0.024459839, 0.0043144226, -0.018188477, 0.030044556, -0.034942627, 0.012260437, -0.035949707, -0.0063438416, -0.043395996, -0.02986145, -0.012130737, ...
http://arxiv.org/abs/2303.05352v2
Extracting Accurate Materials Data from Research Papers with Conversational Language Models and Prompt Engineering
Maciej P. Polak, Dane Morgan
cs.CL, cond-mat.mtrl-sci
There has been a growing effort to replace hand extraction of data from research papers with automated data extraction based on natural language processing, language models, and recently, large language models (LLMs). Although these methods enable efficient extraction of data from large sets of research papers, they re...
http://arxiv.org/pdf/2303.05352v2
2023-06-27T06:59:50
2023-03-07T17:54:53
{ "title": [ -0.07055664, -0.011100769, -0.0635376, -0.007843018, -0.042022705, -0.012878418, -0.06573486, -0.008476257, 0.002193451, 0.024414062, -0.0015392303, -0.0024490356, -0.01285553, -0.034729004, -0.032592773, 0.0068893433, -0.01675415, 0...
http://arxiv.org/abs/2303.04093v1
Marpa and nullable symbols
Jeffrey Kegler
cs.CL
The Marpa parser was intended to make the best results in the academic literature on Earley's algorithm available as a practical general parser. Earley-based parsers have had issues handling nullable symbols. Initially, we dealt with nullable symbols by following the approach in Aycock and Horspool's 2002 paper. This p...
http://arxiv.org/pdf/2303.04093v1
2023-03-07T17:53:00
2023-03-07T17:53:00
{ "title": [ 0.026229858, -0.0001655817, -0.03692627, -0.0038414001, -0.039398193, -0.03439331, -0.015808105, -0.0065841675, 0.01537323, 0.04244995, -0.014457703, -0.039093018, 0.062194824, -0.043273926, 0.03515625, 0.012458801, 0.015335083, 0.01...
http://arxiv.org/abs/2303.04092v1
CroCoSum: A Benchmark Dataset for Cross-Lingual Code-Switched Summarization
Ruochen Zhang, Carsten Eickhoff
cs.CL
Cross-lingual summarization (CLS) has attracted increasing interest in recent years due to the availability of large-scale web-mined datasets and the advancements of multilingual language models. However, given the rareness of naturally occurring CLS resources, the majority of datasets are forced to rely on translation...
http://arxiv.org/pdf/2303.04092v1
2023-03-07T17:52:51
2023-03-07T17:52:51
{ "title": [ -0.030685425, -0.024414062, -0.015792847, -0.05102539, -0.01928711, 0.0124053955, -0.06341553, -0.0032749176, 0.010765076, 0.05239868, -0.012634277, 0.026473999, 0.015533447, 0.0029640198, -0.047210693, 0.02041626, -0.0040740967, 0.0...
http://arxiv.org/abs/2303.04091v3
Abstract Visual Reasoning Enabled by Language
Giacomo Camposampiero, Loic Houmard, Benjamin Estermann, Joël Mathys, Roger Wattenhofer
cs.AI, cs.CL, cs.LG
While artificial intelligence (AI) models have achieved human or even superhuman performance in many well-defined applications, they still struggle to show signs of broad and flexible intelligence. The Abstraction and Reasoning Corpus (ARC), a visual intelligence benchmark introduced by Fran\c{c}ois Chollet, aims to as...
http://arxiv.org/pdf/2303.04091v3
2023-06-22T10:41:41
2023-03-07T17:52:46
{ "title": [ 0.0013475418, -0.013313293, -0.02218628, -0.08190918, -0.0029315948, 0.005302429, -0.03881836, -0.0012254715, 0.03616333, 0.010383606, -0.03488159, -0.031951904, -0.011695862, -0.024810791, -0.040161133, -0.0024719238, 0.023040771, 0...
http://arxiv.org/abs/2303.05392v1
Automatically Summarizing Evidence from Clinical Trials: A Prototype Highlighting Current Challenges
Sanjana Ramprasad, Denis Jered McInerney, Iain J. Marshal, Byron C. Wallace
cs.CL, cs.IR, cs.LG
We present TrialsSummarizer, a system that aims to automatically summarize evidence presented in the set of randomized controlled trials most relevant to a given query. Building on prior work, the system retrieves trial publications matching a query specifying a combination of condition, intervention(s), and outcome(s)...
http://arxiv.org/pdf/2303.05392v1
2023-03-07T17:30:48
2023-03-07T17:30:48
{ "title": [ -0.024169922, -0.003967285, -0.035308838, -0.072021484, -0.008850098, 0.014465332, -0.056488037, -0.007549286, -0.023117065, 0.051361084, -0.009757996, 0.030410767, -0.010269165, -0.0064735413, -0.044891357, 0.013656616, -0.0013093948, ...
http://arxiv.org/abs/2303.04053v3
Describe me an Aucklet: Generating Grounded Perceptual Category Descriptions
Bill Noble, Nikolai Ilinykh
cs.CL
Human speakers can generate descriptions of perceptual concepts, abstracted from the instance-level. Moreover, such descriptions can be used by other speakers to learn provisional representations of those concepts. Learning and using abstract perceptual concepts is under-investigated in the language-and-vision field. T...
http://arxiv.org/pdf/2303.04053v3
2023-10-26T11:35:03
2023-03-07T17:01:25
{ "title": [ -0.006980896, 0.0019350052, -0.042419434, -0.04864502, 0.005680084, -0.024169922, -0.029830933, -0.03945923, 0.0007367134, 0.0289917, 0.0042648315, 0.017318726, 0.0027828217, -0.02468872, -0.043029785, 0.020095825, -0.013885498, 0.05...
http://arxiv.org/abs/2303.04048v3
Is ChatGPT a Good NLG Evaluator? A Preliminary Study
Jiaan Wang, Yunlong Liang, Fandong Meng, Zengkui Sun, Haoxiang Shi, Zhixu Li, Jinan Xu, Jianfeng Qu, Jie Zhou
cs.CL, cs.AI
Recently, the emergence of ChatGPT has attracted wide attention from the computational linguistics community. Many prior studies have shown that ChatGPT achieves remarkable performance on various NLP tasks in terms of automatic evaluation metrics. However, the ability of ChatGPT to serve as an evaluation metric is stil...
http://arxiv.org/pdf/2303.04048v3
2023-10-24T14:56:51
2023-03-07T16:57:20
{ "title": [ -0.009063721, -0.023925781, -0.011482239, -0.025924683, -0.007232666, 0.0037975311, -0.0022792816, -0.009712219, -0.023101807, 0.024108887, -0.012145996, 0.016937256, -0.018463135, -0.0044517517, -0.054534912, 0.0023708344, -0.03643799, ...
http://arxiv.org/abs/2303.05383v1
Making a Computational Attorney
Dell Zhang, Frank Schilder, Jack G. Conrad, Masoud Makrehchi, David von Rickenbach, Isabelle Moulinier
cs.CL, cs.CY, cs.LG, I.2
This "blue sky idea" paper outlines the opportunities and challenges in data mining and machine learning involving making a computational attorney -- an intelligent software agent capable of helping human lawyers with a wide range of complex high-level legal tasks such as drafting legal briefs for the prosecution or de...
http://arxiv.org/pdf/2303.05383v1
2023-03-07T16:44:29
2023-03-07T16:44:29
{ "title": [ -0.009712219, -0.04534912, -0.0027751923, -0.029632568, -0.072143555, 0.012550354, -0.04385376, 0.03036499, -0.0053215027, 0.031188965, -0.043792725, -0.00029802322, 0.0025749207, -0.014579773, 0.0090408325, -0.0069465637, 0.0048103333, ...
http://arxiv.org/abs/2303.04001v2
ELODIN: Naming Concepts in Embedding Spaces
Rodrigo Mello, Filipe Calegario, Geber Ramalho
cs.CV, cs.CL, cs.GR, cs.LG
Despite recent advancements, the field of text-to-image synthesis still suffers from lack of fine-grained control. Using only text, it remains challenging to deal with issues such as concept coherence and concept contamination. We propose a method to enhance control by generating specific concepts that can be reused th...
http://arxiv.org/pdf/2303.04001v2
2023-03-09T17:10:27
2023-03-07T16:00:26
{ "title": [ -0.013885498, -0.00730896, -0.026321411, -0.02558899, 0.0029716492, -0.008659363, 0.0056152344, -0.026138306, -0.003993988, 0.02394104, -0.032440186, 0.025039673, 0.014907837, -0.014251709, 0.00010430813, -0.0058784485, -0.010063171, ...
http://arxiv.org/abs/2303.03975v1
GATE: A Challenge Set for Gender-Ambiguous Translation Examples
Spencer Rarrick, Ranjita Naik, Varun Mathur, Sundar Poudel, Vishal Chowdhary
cs.CL
Although recent years have brought significant progress in improving translation of unambiguously gendered sentences, translation of ambiguously gendered input remains relatively unexplored. When source gender is ambiguous, machine translation models typically default to stereotypical gender roles, perpetuating harmful...
http://arxiv.org/pdf/2303.03975v1
2023-03-07T15:23:38
2023-03-07T15:23:38
{ "title": [ -0.06573486, 0.0007195473, -0.0463562, -0.008636475, -0.008651733, 0.0004916191, -0.025314331, -0.009796143, 0.011566162, 0.041931152, -0.0066108704, -0.0014429092, 0.04736328, -0.027282715, -0.041778564, 0.020019531, -0.02041626, 0....