id
stringlengths
32
39
title
stringlengths
7
232
authors
stringlengths
5
6.98k
categories
stringlengths
12
178
summary
stringlengths
3
3.12k
link_pdf
stringlengths
32
39
updated
timestamp[s]
published
timestamp[s]
embeddings
dict
http://arxiv.org/abs/1612.00377v4
Piecewise Latent Variables for Neural Variational Text Processing
Iulian V. Serban, Alexander G. Ororbia II, Joelle Pineau, Aaron Courville
cs.CL, cs.AI, cs.LG, cs.NE, I.5.1; I.2.7
Advances in neural variational inference have facilitated the learning of powerful directed graphical models with continuous latent variables, such as variational autoencoders. The hope is that such models will learn to represent rich, multi-modal latent factors in real-world data, such as natural language text. Howeve...
http://arxiv.org/pdf/1612.00377v4
2017-09-23T13:33:55
2016-12-01T18:49:23
{ "title": [ -0.029144287, -0.008468628, -0.03857422, -0.047302246, -0.00806427, -0.0014362335, -0.043060303, -0.002412796, 0.00497818, 0.043945312, -0.00868988, 0.009155273, 0.00015091896, -0.0077323914, -0.05987549, 0.004272461, -0.015975952, 0...
http://arxiv.org/abs/1703.03130v1
A Structured Self-attentive Sentence Embedding
Zhouhan Lin, Minwei Feng, Cicero Nogueira dos Santos, Mo Yu, Bing Xiang, Bowen Zhou, Yoshua Bengio
cs.CL, cs.AI, cs.LG, cs.NE
This paper proposes a new model for extracting an interpretable sentence embedding by introducing self-attention. Instead of using a vector, we use a 2-D matrix to represent the embedding, with each row of the matrix attending on a different part of the sentence. We also propose a self-attention mechanism and a special...
http://arxiv.org/pdf/1703.03130v1
2017-03-09T04:42:30
2017-03-09T04:42:30
{ "title": [ 0.0040397644, 0.007835388, -0.005355835, -0.05670166, 0.005104065, -0.0027580261, -0.034942627, -0.006839752, 0.01612854, 0.018356323, -0.01486969, 0.007247925, -0.012779236, -0.011161804, -0.039520264, 0.0070228577, -0.000813961, 0....
http://arxiv.org/abs/1904.05606v1
Multi-lingual Dialogue Act Recognition with Deep Learning Methods
Jiří Martínek, Pavel Král, Ladislav Lenc, Christophe Cerisara
cs.CL, cs.AI, cs.LG, cs.NE
This paper deals with multi-lingual dialogue act (DA) recognition. The proposed approaches are based on deep neural networks and use word2vec embeddings for word representation. Two multi-lingual models are proposed for this task. The first approach uses one general model trained on the embeddings from all available la...
http://arxiv.org/pdf/1904.05606v1
2019-04-11T09:55:41
2019-04-11T09:55:41
{ "title": [ -0.008834839, 0.0034446716, -0.057678223, -0.06518555, 0.012840271, -0.0154800415, -0.053222656, 0.014305115, 0.015197754, 0.025527954, 0.001452446, -0.018341064, -0.010826111, -0.030822754, -0.056365967, -0.001666069, -0.011955261, ...
http://arxiv.org/abs/1906.00499v1
Budgeted Policy Learning for Task-Oriented Dialogue Systems
Zhirui Zhang, Xiujun Li, Jianfeng Gao, Enhong Chen
cs.CL, cs.AI, cs.LG, cs.NE
This paper presents a new approach that extends Deep Dyna-Q (DDQ) by incorporating a Budget-Conscious Scheduling (BCS) to best utilize a fixed, small amount of user interactions (budget) for learning task-oriented dialogue agents. BCS consists of (1) a Poisson-based global scheduler to allocate budget over different st...
http://arxiv.org/pdf/1906.00499v1
2019-06-02T22:53:33
2019-06-02T22:53:33
{ "title": [ -0.0030670166, -0.011886597, -0.045074463, -0.06555176, -0.0034828186, -0.017044067, -0.029220581, 0.02696228, -0.029769897, 0.016921997, -0.024093628, 0.01424408, -0.03717041, -0.02053833, -0.024459839, -0.031219482, -0.00982666, 0....
http://arxiv.org/abs/1710.00641v1
Improving speech recognition by revising gated recurrent units
Mirco Ravanelli, Philemon Brakel, Maurizio Omologo, Yoshua Bengio
cs.CL, cs.AI, cs.LG, cs.NE
Speech recognition is largely taking advantage of deep learning, showing that substantial benefits can be obtained by modern Recurrent Neural Networks (RNNs). The most popular RNNs are Long Short-Term Memory (LSTMs), which typically reach state-of-the-art performance in many tasks thanks to their ability to learn long-...
http://arxiv.org/pdf/1710.00641v1
2017-09-29T12:40:50
2017-09-29T12:40:50
{ "title": [ -0.015197754, 0.00806427, -0.057647705, -0.060455322, 0.021118164, -0.011024475, -0.03366089, 0.024093628, -0.019256592, 0.05618286, -0.027328491, 0.006919861, 0.0059127808, 0.004337311, -0.06347656, 0.0020332336, -0.01676941, 0.0545...
http://arxiv.org/abs/1802.07089v2
Attentive Tensor Product Learning
Qiuyuan Huang, Li Deng, Dapeng Wu, Chang Liu, Xiaodong He
cs.CL, cs.AI, cs.LG, cs.NE
This paper proposes a new architecture - Attentive Tensor Product Learning (ATPL) - to represent grammatical structures in deep learning models. ATPL is a new architecture to bridge this gap by exploiting Tensor Product Representations (TPR), a structured neural-symbolic model developed in cognitive science, aiming to ...
http://arxiv.org/pdf/1802.07089v2
2018-11-01T05:14:18
2018-02-20T12:42:07
{ "title": [ 0.002084732, 0.016830444, -0.03643799, -0.04623413, 0.022521973, -0.014137268, -0.030776978, -0.036621094, -0.0039863586, 0.034942627, -0.01828003, -0.004802704, -0.00013530254, -0.034088135, -0.019165039, 0.0113220215, 0.007369995, ...
http://arxiv.org/abs/2004.14870v4
Mind Your Inflections! Improving NLP for Non-Standard Englishes with Base-Inflection Encoding
Samson Tan, Shafiq Joty, Lav R. Varshney, Min-Yen Kan
cs.CL, cs.AI, cs.LG, cs.NE
Inflectional variation is a common feature of World Englishes such as Colloquial Singapore English and African American Vernacular English. Although comprehension by human readers is usually unimpaired by non-standard inflections, current NLP systems are not yet robust. We propose Base-Inflection Encoding (BITE), a met...
http://arxiv.org/pdf/2004.14870v4
2020-11-18T06:16:31
2020-04-30T15:15:40
{ "title": [ -0.033325195, -0.008407593, -0.055786133, -0.018447876, 0.007171631, -0.028396606, -0.005302429, 0.007663727, -0.0050468445, 0.032928467, -0.035217285, 0.033843994, 0.01977539, -0.018005371, -0.007205963, 0.007003784, -0.020706177, 0...
http://arxiv.org/abs/1509.00838v2
What to talk about and how? Selective Generation using LSTMs with Coarse-to-Fine Alignment
Hongyuan Mei, Mohit Bansal, Matthew R. Walter
cs.CL, cs.AI, cs.LG, cs.NE
We propose an end-to-end, domain-independent neural encoder-aligner-decoder model for selective generation, i.e., the joint task of content selection and surface realization. Our model first encodes a full set of over-determined database event records via an LSTM-based recurrent neural network, then utilizes a novel co...
http://arxiv.org/pdf/1509.00838v2
2016-01-08T23:07:32
2015-09-02T19:52:56
{ "title": [ -0.007762909, 0.0118255615, -0.043151855, -0.056030273, 0.011276245, -0.009178162, -0.046417236, -0.007881165, -0.01727295, 0.021652222, -0.002960205, 0.000076413155, -0.007949829, -0.04547119, -0.008079529, 0.0036411285, -0.00983429, ...
http://arxiv.org/abs/1805.12352v2
DialogWAE: Multimodal Response Generation with Conditional Wasserstein Auto-Encoder
Xiaodong Gu, Kyunghyun Cho, Jung-Woo Ha, Sunghun Kim
cs.CL, cs.AI, cs.LG, cs.NE
Variational autoencoders~(VAEs) have shown a promise in data-driven conversation modeling. However, most VAE conversation models match the approximate posterior distribution over the latent variables to a simple prior such as standard normal distribution, thereby restricting the generated responses to a relatively simp...
http://arxiv.org/pdf/1805.12352v2
2019-02-26T02:32:44
2018-05-31T07:25:04
{ "title": [ -0.026763916, -0.00472641, -0.05532837, -0.06463623, -0.0104522705, 0.0014743805, -0.074035645, 0.0048713684, 0.0006661415, 0.03475952, 0.007156372, 0.0051841736, -0.0005350113, -0.017562866, -0.041381836, 0.004798889, -0.002412796, ...
http://arxiv.org/abs/1811.00147v1
DOLORES: Deep Contextualized Knowledge Graph Embeddings
Haoyu Wang, Vivek Kulkarni, William Yang Wang
cs.CL, cs.AI, cs.LG, cs.NE
We introduce a new method DOLORES for learning knowledge graph embeddings that effectively captures contextual cues and dependencies among entities and relations. First, we note that short paths on knowledge graphs comprising of chains of entities and relations can encode valuable information regarding their contextual...
http://arxiv.org/pdf/1811.00147v1
2018-10-31T22:59:57
2018-10-31T22:59:57
{ "title": [ -0.021453857, -0.007709503, -0.06237793, -0.0848999, 0.0066833496, 0.0023021698, -0.06500244, 0.00970459, 0.007232666, 0.02784729, -0.029724121, 0.02722168, 0.0073661804, -0.00283432, -0.025390625, 0.006439209, -0.007472992, 0.042449...
http://arxiv.org/abs/1811.00196v2
Towards Explainable NLP: A Generative Explanation Framework for Text Classification
Hui Liu, Qingyu Yin, William Yang Wang
cs.CL, cs.AI, cs.LG, cs.NE
Building explainable systems is a critical problem in the field of Natural Language Processing (NLP), since most machine learning models provide no explanations for the predictions. Existing approaches for explainable machine learning systems tend to focus on interpreting the outputs or the connections between inputs a...
http://arxiv.org/pdf/1811.00196v2
2019-06-11T13:12:58
2018-11-01T02:45:57
{ "title": [ -0.024978638, -0.010246277, -0.017410278, -0.06335449, -0.009056091, 0.016784668, -0.044036865, -0.0065727234, 0.008460999, -0.005493164, -0.022781372, 0.016342163, -0.035125732, -0.010528564, -0.036193848, -0.01663208, -0.035614014, ...
http://arxiv.org/abs/1811.00198v1
MOHONE: Modeling Higher Order Network Effects in KnowledgeGraphs via Network Infused Embeddings
Hao Yu, Vivek Kulkarni, William Wang
cs.CL, cs.AI, cs.LG, cs.NE
Many knowledge graph embedding methods operate on triples and are therefore implicitly limited by a very local view of the entire knowledge graph. We present a new framework MOHONE to effectively model higher order network effects in knowledge-graphs, thus enabling one to capture varying degrees of network connectivity...
http://arxiv.org/pdf/1811.00198v1
2018-11-01T03:04:09
2018-11-01T03:04:09
{ "title": [ -0.023880005, -0.010864258, -0.06311035, -0.0647583, -0.008766174, -0.008743286, -0.04748535, 0.0038604736, 0.00015234947, 0.019638062, -0.042022705, 0.01852417, -0.0033035278, 0.00013411045, -0.03237915, -0.014015198, -0.008872986, ...
http://arxiv.org/abs/1811.07550v1
Switch-based Active Deep Dyna-Q: Efficient Adaptive Planning for Task-Completion Dialogue Policy Learning
Yuexin Wu, Xiujun Li, Jingjing Liu, Jianfeng Gao, Yiming Yang
cs.CL, cs.AI, cs.LG, cs.NE
Training task-completion dialogue agents with reinforcement learning usually requires a large number of real user experiences. The Dyna-Q algorithm extends Q-learning by integrating a world model, and thus can effectively boost training efficiency using simulated experiences generated by the world model. The effectiven...
http://arxiv.org/pdf/1811.07550v1
2018-11-19T08:23:34
2018-11-19T08:23:34
{ "title": [ 0.008430481, 0.0023231506, -0.03286743, -0.07196045, 0.017730713, -0.0072021484, -0.030838013, 0.006641388, -0.024871826, 0.0104904175, -0.013076782, 0.0068969727, -0.018966675, -0.042816162, 0.008728027, -0.023040771, -0.0017595291, ...
http://arxiv.org/abs/1901.06610v2
Hierarchical Attentional Hybrid Neural Networks for Document Classification
Jader Abreu, Luis Fred, David Macêdo, Cleber Zanchettin
cs.CL, cs.AI, cs.LG, cs.NE
Document classification is a challenging task with important applications. The deep learning approaches to the problem have gained much attention recently. Despite the progress, the proposed models do not incorporate the knowledge of the document structure in the architecture efficiently and not take into account the c...
http://arxiv.org/pdf/1901.06610v2
2019-06-28T17:49:18
2019-01-20T01:48:43
{ "title": [ -0.011764526, 0.017974854, -0.03677368, -0.046081543, 0.01361084, -0.013549805, -0.025802612, -0.025100708, 0.018463135, 0.046020508, -0.04272461, -0.0059928894, 0.025817871, 0.0024204254, -0.05731201, -0.0033397675, -0.016616821, 0....
http://arxiv.org/abs/1909.00141v2
Deep Reinforcement Learning with Distributional Semantic Rewards for Abstractive Summarization
Siyao Li, Deren Lei, Pengda Qin, William Yang Wang
cs.CL, cs.AI, cs.LG, cs.NE
Deep reinforcement learning (RL) has been a commonly-used strategy for the abstractive summarization task to address both the exposure bias and non-differentiable task issues. However, the conventional reward Rouge-L simply looks for exact n-grams matches between candidates and annotated references, which inevitably ma...
http://arxiv.org/pdf/1909.00141v2
2019-09-10T23:30:26
2019-08-31T06:13:33
{ "title": [ 0.0041503906, 0.011619568, -0.033294678, -0.06762695, 0.0037345886, -0.006832123, -0.04244995, -0.016784668, -0.008659363, 0.03656006, -0.016281128, 0.027618408, -0.03289795, -0.008613586, -0.031143188, -0.007865906, -0.005218506, 0....
http://arxiv.org/abs/2010.12681v1
Robust Document Representations using Latent Topics and Metadata
Natraj Raman, Armineh Nourbakhsh, Sameena Shah, Manuela Veloso
cs.CL, cs.AI, cs.LG, cs.NE, I.2.7; I.7.0
Task specific fine-tuning of a pre-trained neural language model using a custom softmax output layer is the de facto approach of late when dealing with document classification problems. This technique is not adequate when labeled examples are not available at training time and when the metadata artifacts in a document ...
http://arxiv.org/pdf/2010.12681v1
2020-10-23T21:52:38
2020-10-23T21:52:38
{ "title": [ -0.0012760162, 0.011177063, -0.056640625, -0.06085205, 0.023956299, -0.015731812, -0.041015625, -0.023757935, 0.018951416, 0.031433105, -0.036987305, 0.035461426, -0.004371643, -0.002729416, -0.030593872, 0.012176514, 0.0030498505, 0...
http://arxiv.org/abs/2011.10208v1
Collaborative Storytelling with Large-scale Neural Language Models
Eric Nichols, Leo Gao, Randy Gomez
cs.CL, cs.AI, cs.LG, cs.NE
Storytelling plays a central role in human socializing and entertainment. However, much of the research on automatic storytelling generation assumes that stories will be generated by an agent without any human interaction. In this paper, we introduce the task of collaborative storytelling, where an artificial intellige...
http://arxiv.org/pdf/2011.10208v1
2020-11-20T04:36:54
2020-11-20T04:36:54
{ "title": [ -0.041748047, -0.008430481, -0.051818848, -0.04736328, -0.0078048706, -0.0011463165, -0.047698975, 0.012992859, -0.009056091, 0.027511597, 0.009468079, -0.011978149, -0.017303467, -0.022705078, -0.034118652, -0.002632141, -0.026107788, ...
http://arxiv.org/abs/2012.09823v1
Continual Lifelong Learning in Natural Language Processing: A Survey
Magdalena Biesialska, Katarzyna Biesialska, Marta R. Costa-jussà
cs.CL, cs.AI, cs.LG, cs.NE
Continual learning (CL) aims to enable information systems to learn from a continuous data stream across time. However, it is difficult for existing deep learning architectures to learn a new task without largely forgetting previously acquired knowledge. Furthermore, CL is particularly challenging for language learning...
http://arxiv.org/pdf/2012.09823v1
2020-12-17T18:44:36
2020-12-17T18:44:36
{ "title": [ -0.0012989044, -0.0007472038, -0.0362854, -0.078430176, 0.017913818, -0.011001587, -0.02532959, 0.010238647, 0.012863159, 0.03086853, -0.027877808, 0.020751953, -0.01626587, -0.0049362183, -0.051086426, 0.0026016235, -0.009498596, 0....
http://arxiv.org/abs/2108.08708v3
Czech News Dataset for Semantic Textual Similarity
Jakub Sido, Michal Seják, Ondřej Pražák, Miloslav Konopík, Václav Moravec
cs.CL, cs.AI, cs.LG, cs.NE
This paper describes a novel dataset consisting of sentences with semantic similarity annotations. The data originate from the journalistic domain in the Czech language. We describe the process of collecting and annotating the data in detail. The dataset contains 138,556 human annotations divided into train and test se...
http://arxiv.org/pdf/2108.08708v3
2022-01-21T10:28:54
2021-08-19T14:20:17
{ "title": [ -0.04043579, -0.00065374374, -0.058746338, -0.04824829, 0.016189575, -0.0011482239, -0.026809692, -0.004131317, -0.0007753372, 0.064453125, -0.029220581, 0.026229858, 0.0390625, 0.0034866333, -0.03756714, 0.025421143, -0.006160736, 0...
http://arxiv.org/abs/2110.01804v1
A Survey On Neural Word Embeddings
Erhan Sezerer, Selma Tekir
cs.CL, cs.AI, cs.LG, cs.NE
Understanding human language has been a sub-challenge on the way of intelligent machines. The study of meaning in natural language processing (NLP) relies on the distributional hypothesis where language elements get meaning from the words that co-occur within contexts. The revolutionary idea of distributed representati...
http://arxiv.org/pdf/2110.01804v1
2021-10-05T03:37:57
2021-10-05T03:37:57
{ "title": [ -0.015556335, -0.0034103394, -0.027618408, -0.039367676, 0.0073165894, 0.0032787323, -0.029647827, -0.0036296844, 0.017730713, 0.034088135, -0.013038635, 0.013198853, 0.0046463013, -0.005252838, -0.040802002, 0.021209717, -0.011451721, ...
http://arxiv.org/abs/2110.03727v2
Contextual Sentence Classification: Detecting Sustainability Initiatives in Company Reports
Dan Hirlea, Christopher Bryant, Maurizio Zollo, Marek Rei
cs.CL, cs.AI, cs.LG, cs.NE
We introduce the novel task of detecting sustainability initiatives in company reports. Given a full report, the aim is to automatically identify mentions of practical activities that a company has performed in order to tackle specific societal issues. New methods for identifying continuous sentence spans need to be de...
http://arxiv.org/pdf/2110.03727v2
2022-03-03T21:42:06
2021-10-07T18:28:03
{ "title": [ 0.0016946793, 0.028457642, -0.04663086, -0.05267334, 0.023895264, -0.010429382, -0.0317688, -0.021697998, -0.010314941, 0.026321411, -0.053771973, -0.0024528503, -0.009529114, 0.0037441254, -0.046142578, -0.04864502, -0.047180176, 0....
http://arxiv.org/abs/2112.02706v1
Achieving Forgetting Prevention and Knowledge Transfer in Continual Learning
Zixuan Ke, Bing Liu, Nianzu Ma, Hu Xu, Lei Shu
cs.CL, cs.AI, cs.LG, cs.NE
Continual learning (CL) learns a sequence of tasks incrementally with the goal of achieving two main objectives: overcoming catastrophic forgetting (CF) and encouraging knowledge transfer (KT) across tasks. However, most existing techniques focus only on overcoming CF and have no mechanism to encourage KT, and thus do ...
http://arxiv.org/pdf/2112.02706v1
2021-12-05T23:13:13
2021-12-05T23:13:13
{ "title": [ -0.0017032623, 0.015197754, -0.064208984, -0.076538086, 0.035247803, -0.03289795, -0.015853882, 0.0023651123, -0.024459839, 0.03314209, 0.00724411, 0.008514404, -0.026473999, -0.028045654, -0.047790527, 0.0006814003, 0.013504028, 0.0...
http://arxiv.org/abs/2112.02714v1
CLASSIC: Continual and Contrastive Learning of Aspect Sentiment Classification Tasks
Zixuan Ke, Bing Liu, Hu Xu, Lei Shu
cs.CL, cs.AI, cs.LG, cs.NE
This paper studies continual learning (CL) of a sequence of aspect sentiment classification(ASC) tasks in a particular CL setting called domain incremental learning (DIL). Each task is from a different domain or product. The DIL setting is particularly suited to ASC because in testing the system needs not know the task...
http://arxiv.org/pdf/2112.02714v1
2021-12-05T23:55:53
2021-12-05T23:55:53
{ "title": [ 0.01637268, 0.0034999847, -0.028533936, -0.05407715, 0.0141067505, -0.0231781, -0.016159058, -0.01008606, 0.037109375, 0.039764404, -0.025177002, -0.007881165, 0.006969452, -0.01473999, -0.07067871, 0.017730713, 0.00043010712, 0.0549...
http://arxiv.org/abs/2112.03271v1
Adapting BERT for Continual Learning of a Sequence of Aspect Sentiment Classification Tasks
Zixuan Ke, Hu Xu, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
This paper studies continual learning (CL) of a sequence of aspect sentiment classification (ASC) tasks. Although some CL techniques have been proposed for document sentiment classification, we are not aware of any CL work on ASC. A CL system that incrementally learns a sequence of ASC tasks should address the followin...
http://arxiv.org/pdf/2112.03271v1
2021-12-06T02:46:06
2021-12-06T02:46:06
{ "title": [ 0.00932312, -0.009307861, -0.015060425, -0.05834961, -0.005203247, -0.0058784485, -0.034057617, 0.0046806335, 0.037994385, 0.04586792, -0.028640747, -0.013465881, 0.00699234, -0.0014381409, -0.09069824, 0.008903503, -0.008674622, 0.0...
http://arxiv.org/abs/2112.10021v1
Continual Learning with Knowledge Transfer for Sentiment Classification
Zixuan Ke, Bing Liu, Hao Wang, Lei Shu
cs.CL, cs.AI, cs.LG, cs.NE
This paper studies continual learning (CL) for sentiment classification (SC). In this setting, the CL system learns a sequence of SC tasks incrementally in a neural network, where each task builds a classifier to classify the sentiment of reviews of a particular product category or domain. Two natural questions are: Ca...
http://arxiv.org/pdf/2112.10021v1
2021-12-18T22:58:21
2021-12-18T22:58:21
{ "title": [ 0.0023384094, 0.0063591003, -0.032470703, -0.04586792, 0.016357422, -0.020935059, -0.011436462, -0.015533447, 0.023590088, 0.03161621, -0.033599854, -0.0043945312, -0.013969421, -0.0035686493, -0.072143555, -0.0073394775, 0.0029640198, ...
http://arxiv.org/abs/2206.01512v1
Latent Topology Induction for Understanding Contextualized Representations
Yao Fu, Mirella Lapata
cs.CL, cs.AI, cs.LG, cs.NE
In this work, we study the representation space of contextualized embeddings and gain insight into the hidden topology of large language models. We show there exists a network of latent states that summarize linguistic properties of contextualized representations. Instead of seeking alignments to existing well-defined ...
http://arxiv.org/pdf/2206.01512v1
2022-06-03T11:22:48
2022-06-03T11:22:48
{ "title": [ 0.02923584, 0.0055122375, -0.03527832, -0.06439209, 0.011627197, -0.024856567, -0.01826477, -0.012275696, 0.012252808, 0.021102905, -0.04663086, 0.023605347, -0.023666382, -0.014343262, -0.014022827, -0.0017089844, 0.00819397, 0.0151...
http://arxiv.org/abs/2208.05720v1
A Model of Anaphoric Ambiguities using Sheaf Theoretic Quantum-like Contextuality and BERT
Kin Ian Lo, Mehrnoosh Sadrzadeh, Shane Mansfield
cs.CL, cs.AI, cs.LG, cs.NE
Ambiguities of natural language do not preclude us from using it and context helps in getting ideas across. They, nonetheless, pose a key challenge to the development of competent machines to understand natural language and use it as humans do. Contextuality is an unparalleled phenomenon in quantum mechanics, where dif...
http://arxiv.org/pdf/2208.05720v1
2022-08-11T09:31:15
2022-08-11T09:31:15
{ "title": [ -0.004764557, 0.019454956, -0.022644043, -0.028900146, -0.010139465, -0.014404297, -0.03488159, -0.029663086, -0.014564514, 0.077697754, 0.024215698, -0.0018138885, 0.022064209, -0.021987915, -0.056121826, 0.015342712, -0.031402588, ...
http://arxiv.org/abs/2210.05549v1
Continual Training of Language Models for Few-Shot Learning
Zixuan Ke, Haowei Lin, Yijia Shao, Hu Xu, Lei Shu, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
Recent work on applying large language models (LMs) achieves impressive performance in many NLP applications. Adapting or posttraining an LM using an unlabeled domain corpus can produce even better performance for end-tasks in the domain. This paper proposes the problem of continually extending an LM by incrementally p...
http://arxiv.org/pdf/2210.05549v1
2022-10-11T15:43:58
2022-10-11T15:43:58
{ "title": [ -0.014129639, -0.021759033, -0.022033691, -0.041900635, -0.017745972, 0.000007390976, -0.043914795, 0.01159668, -0.00092220306, 0.02909851, -0.021438599, 0.008323669, -0.010040283, -0.015449524, -0.04333496, 0.0016021729, 0.0017786026, ...
http://arxiv.org/abs/2211.00582v1
ClassActionPrediction: A Challenging Benchmark for Legal Judgment Prediction of Class Action Cases in the US
Gil Semo, Dor Bernsohn, Ben Hagag, Gila Hayat, Joel Niklaus
cs.CL, cs.AI, cs.LG, cs.NE, 68T50, I.2; I.7
The research field of Legal Natural Language Processing (NLP) has been very active recently, with Legal Judgment Prediction (LJP) becoming one of the most extensively studied tasks. To date, most publicly released LJP datasets originate from countries with civil law. In this work, we release, for the first time, a chal...
http://arxiv.org/pdf/2211.00582v1
2022-11-01T16:57:59
2022-11-01T16:57:59
{ "title": [ 0.0053100586, -0.0023860931, -0.04748535, -0.06323242, 0.00472641, -0.030715942, -0.012626648, 0.003610611, 0.01360321, 0.026641846, -0.060791016, 0.029541016, 0.009498596, 0.0011663437, -0.024993896, -0.009559631, -0.0010681152, 0.0...
http://arxiv.org/abs/2211.12701v2
Continual Learning of Natural Language Processing Tasks: A Survey
Zixuan Ke, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
Continual learning (CL) is a learning paradigm that emulates the human capability of learning and accumulating knowledge continually without forgetting the previously learned knowledge and also transferring the learned knowledge to help learn new tasks better. This survey presents a comprehensive review and analysis of...
http://arxiv.org/pdf/2211.12701v2
2023-05-11T09:48:55
2022-11-23T04:46:28
{ "title": [ -0.007843018, 0.0011806488, -0.032348633, -0.060699463, 0.016921997, -0.011276245, -0.029876709, 0.00504303, 0.0137786865, 0.024215698, -0.031021118, 0.013664246, -0.018249512, -0.0036087036, -0.05206299, 0.0037631989, -0.012420654, ...
http://arxiv.org/abs/2301.08986v1
Adapting a Language Model While Preserving its General Knowledge
Zixuan Ke, Yijia Shao, Haowei Lin, Hu Xu, Lei Shu, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
Domain-adaptive pre-training (or DA-training for short), also known as post-training, aims to train a pre-trained general-purpose language model (LM) using an unlabeled corpus of a particular domain to adapt the LM so that end-tasks in the domain can give improved performances. However, existing DA-training methods are...
http://arxiv.org/pdf/2301.08986v1
2023-01-21T17:57:53
2023-01-21T17:57:53
{ "title": [ -0.016616821, -0.02142334, -0.04055786, -0.07232666, -0.042388916, -0.005191803, -0.07122803, 0.025878906, 0.02330017, 0.049957275, -0.021621704, -0.029434204, 0.021728516, -0.023712158, -0.048919678, 0.012199402, 0.027236938, -0.002...
http://arxiv.org/abs/2302.12840v2
HULAT at SemEval-2023 Task 10: Data augmentation for pre-trained transformers applied to the detection of sexism in social media
Isabel Segura-Bedmar
cs.CL, cs.AI, cs.LG, cs.NE
This paper describes our participation in SemEval-2023 Task 10, whose goal is the detection of sexism in social media. We explore some of the most popular transformer models such as BERT, DistilBERT, RoBERTa, and XLNet. We also study different data augmentation techniques to increase the training dataset. During the de...
http://arxiv.org/pdf/2302.12840v2
2023-03-01T08:43:13
2023-02-24T18:17:38
{ "title": [ 0.04864502, 0.009941101, -0.03277588, -0.025024414, 0.010032654, -0.037963867, 0.009513855, -0.011634827, -0.0011091232, 0.04916382, -0.039764404, 0.05029297, 0.037719727, -0.0007548332, -0.0017957687, 0.0013246536, -0.017913818, 0.0...
http://arxiv.org/abs/2303.16760v1
ACO-tagger: A Novel Method for Part-of-Speech Tagging using Ant Colony Optimization
Amirhossein Mohammadi, Sara Hajiaghajani, Mohammad Bahrani
cs.CL, cs.AI, cs.LG, cs.NE
Swarm Intelligence algorithms have gained significant attention in recent years as a means of solving complex and non-deterministic problems. These algorithms are inspired by the collective behavior of natural creatures, and they simulate this behavior to develop intelligent agents for computational tasks. One such alg...
http://arxiv.org/pdf/2303.16760v1
2023-03-27T11:48:40
2023-03-27T11:48:40
{ "title": [ -0.02180481, 0.0045814514, -0.027236938, -0.053833008, 0.0135269165, 0.009376526, -0.046295166, -0.0107421875, -0.042907715, 0.062316895, 0.0006785393, 0.048034668, 0.004650116, -0.018310547, -0.024765015, 0.013587952, -0.022323608, ...
http://arxiv.org/abs/2304.14339v1
MarsEclipse at SemEval-2023 Task 3: Multi-Lingual and Multi-Label Framing Detection with Contrastive Learning
Qisheng Liao, Meiting Lai, Preslav Nakov
cs.CL, cs.AI, cs.LG, cs.NE, 68T50, F.2.2; I.2.7
This paper describes our system for SemEval-2023 Task 3 Subtask 2 on Framing Detection. We used a multi-label contrastive loss for fine-tuning large pre-trained language models in a multi-lingual setting, achieving very competitive results: our system was ranked first on the official test set and on the official shared...
http://arxiv.org/pdf/2304.14339v1
2023-04-20T18:42:23
2023-04-20T18:42:23
{ "title": [ 0.018615723, -0.0016365051, -0.018325806, -0.026626587, 0.0005226135, -0.031951904, -0.025619507, -0.0340271, -0.0057373047, 0.028915405, -0.02519226, 0.015823364, 0.006511688, -0.025863647, -0.016052246, 0.0027713776, -0.020355225, ...
http://arxiv.org/abs/2305.11070v1
Enriching language models with graph-based context information to better understand textual data
Albert Roethel, Maria Ganzha, Anna Wróblewska
cs.CL, cs.AI, cs.LG, cs.NE
A considerable number of texts encountered daily are somehow connected with each other. For example, Wikipedia articles refer to other articles via hyperlinks, scientific papers relate to others via citations or (co)authors, while tweets relate via users that follow each other or reshare content. Hence, a graph-like st...
http://arxiv.org/pdf/2305.11070v1
2023-05-10T10:57:21
2023-05-10T10:57:21
{ "title": [ -0.02911377, -0.01499939, -0.047088623, -0.083618164, -0.0020160675, 0.003894806, -0.062347412, 0.022659302, 0.009109497, 0.033294678, -0.02722168, 0.009811401, 0.009902954, -0.0014953613, -0.042633057, 0.005672455, -0.01411438, 0.04...
http://arxiv.org/abs/2306.12619v2
Class-Incremental Learning based on Label Generation
Yijia Shao, Yiduo Guo, Dongyan Zhao, Bing Liu
cs.CL, cs.AI, cs.LG, cs.NE
Despite the great success of pre-trained language models, it is still a challenge to use these models for continual learning, especially for the class-incremental learning (CIL) setting due to catastrophic forgetting (CF). This paper reports our finding that if we formulate CIL as a continual label generation problem, ...
http://arxiv.org/pdf/2306.12619v2
2023-07-20T16:04:19
2023-06-22T01:14:47
{ "title": [ -0.023498535, -0.011795044, -0.056274414, -0.051116943, 0.009986877, -0.02029419, -0.031707764, -0.0027065277, 0.0064353943, 0.004600525, -0.038482666, 0.021896362, -0.036193848, -0.010658264, -0.02406311, -0.009475708, -0.0054473877, ...
http://arxiv.org/abs/2306.13840v2
Beyond Scale: the Diversity Coefficient as a Data Quality Metric Demonstrates LLMs are Pre-trained on Formally Diverse Data
Alycia Lee, Brando Miranda, Sudharsan Sundar, Sanmi Koyejo
cs.CL, cs.AI, cs.LG, cs.NE
Current trends to pre-train capable Large Language Models (LLMs) mostly focus on scaling of model and dataset size. However, the quality of pre-training data is an important factor for training powerful LLMs, yet it is a nebulous concept that has not been fully characterized. Therefore, we use the recently proposed Tas...
http://arxiv.org/pdf/2306.13840v2
2023-09-26T23:29:05
2023-06-24T02:25:56
{ "title": [ -0.004055023, 0.018844604, -0.052947998, -0.03805542, -0.0022716522, -0.031051636, -0.051239014, -0.010810852, -0.026245117, -0.00091028214, -0.038909912, 0.010314941, -0.029693604, -0.035614014, 0.01991272, -0.05734253, -0.00705719, ...
http://arxiv.org/abs/2309.16797v1
Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution
Chrisantha Fernando, Dylan Banarse, Henryk Michalewski, Simon Osindero, Tim Rocktäschel
cs.CL, cs.AI, cs.LG, cs.NE
Popular prompt strategies like Chain-of-Thought Prompting can dramatically improve the reasoning abilities of Large Language Models (LLMs) in various domains. However, such hand-crafted prompt-strategies are often sub-optimal. In this paper, we present Promptbreeder, a general-purpose self-referential self-improvement ...
http://arxiv.org/pdf/2309.16797v1
2023-09-28T19:01:07
2023-09-28T19:01:07
{ "title": [ -0.0061454773, -0.004852295, -0.037322998, -0.038024902, 0.0057754517, -0.014701843, 0.0045318604, -0.0054512024, -0.025543213, 0.023956299, 0.01473999, 0.016403198, -0.016235352, -0.040863037, -0.013656616, 0.018157959, 0.007785797, ...
http://arxiv.org/abs/2310.09436v1
Sub-network Discovery and Soft-masking for Continual Learning of Mixed Tasks
Zixuan Ke, Bing Liu, Wenhan Xiong, Asli Celikyilmaz, Haoran Li
cs.CL, cs.AI, cs.LG, cs.NE
Continual learning (CL) has two main objectives: preventing catastrophic forgetting (CF) and encouraging knowledge transfer (KT). The existing literature mainly focused on overcoming CF. Some work has also been done on KT when the tasks are similar. To our knowledge, only one method has been proposed to learn a sequenc...
http://arxiv.org/pdf/2310.09436v1
2023-10-13T23:00:39
2023-10-13T23:00:39
{ "title": [ -0.06210327, 0.009277344, -0.07635498, -0.0073776245, -0.0050315857, -0.036224365, -0.011993408, 0.012229919, -0.043060303, 0.028381348, -0.020065308, 0.018508911, -0.011001587, -0.033416748, 0.0032157898, -0.01902771, 0.005241394, 0...
http://arxiv.org/abs/2007.04979v1
A Cordial Sync: Going Beyond Marginal Policies for Multi-Agent Embodied Tasks
Unnat Jain, Luca Weihs, Eric Kolve, Ali Farhadi, Svetlana Lazebnik, Aniruddha Kembhavi, Alexander Schwing
cs.CV, cs.AI, cs.LG, cs.MA
Autonomous agents must learn to collaborate. It is not scalable to develop a new centralized agent every time a task's difficulty outpaces a single agent's abilities. While multi-agent collaboration research has flourished in gridworld-like environments, relatively little work has considered visually rich domains. Addr...
http://arxiv.org/pdf/2007.04979v1
2020-07-09T17:59:57
2020-07-09T17:59:57
{ "title": [ -0.024093628, 0.004169464, -0.0071487427, -0.025863647, -0.0005979538, -0.0036525726, -0.001660347, -0.023208618, -0.026229858, 0.021713257, -0.026245117, 0.025100708, -0.021347046, 0.0005478859, -0.03100586, -0.019805908, -0.008850098, ...
http://arxiv.org/abs/2106.02067v2
Learning to Draw: Emergent Communication through Sketching
Daniela Mihai, Jonathon Hare
cs.CV, cs.AI, cs.LG, cs.MA
Evidence that visual communication preceded written language and provided a basis for it goes back to prehistory, in forms such as cave and rock paintings depicting traces of our distant ancestors. Emergent communication research has sought to explore how agents can learn to communicate in order to collaboratively solv...
http://arxiv.org/pdf/2106.02067v2
2021-11-09T16:35:33
2021-06-03T18:17:55
{ "title": [ -0.035247803, -0.009925842, -0.014564514, 0.012382507, -0.015075684, -0.008117676, 0.023452759, -0.032409668, -0.008323669, 0.019104004, -0.023757935, 0.0057411194, -0.020233154, -0.020706177, -0.034698486, -0.02027893, 0.0024414062, ...
http://arxiv.org/abs/2110.05769v1
Interpretation of Emergent Communication in Heterogeneous Collaborative Embodied Agents
Shivansh Patel, Saim Wani, Unnat Jain, Alexander Schwing, Svetlana Lazebnik, Manolis Savva, Angel X. Chang
cs.CV, cs.AI, cs.LG, cs.MA
Communication between embodied AI agents has received increasing attention in recent years. Despite its use, it is still unclear whether the learned communication is interpretable and grounded in perception. To study the grounding of emergent forms of communication, we first introduce the collaborative multi-object nav...
http://arxiv.org/pdf/2110.05769v1
2021-10-12T06:56:11
2021-10-12T06:56:11
{ "title": [ -0.047088623, 0.013946533, -0.021911621, -0.0072898865, -0.0014476776, -0.0031547546, -0.011741638, -0.027954102, -0.0035190582, 0.031463623, -0.01574707, 0.0056114197, 0.0021400452, -0.010757446, -0.04324341, -0.016860962, -0.016723633,...
http://arxiv.org/abs/1804.03980v1
Emergent Communication through Negotiation
Kris Cao, Angeliki Lazaridou, Marc Lanctot, Joel Z Leibo, Karl Tuyls, Stephen Clark
cs.AI, cs.CL, cs.LG, cs.MA
Multi-agent reinforcement learning offers a way to study how communication could emerge in communities of agents needing to solve specific problems. In this paper, we study the emergence of communication in the negotiation environment, a semi-cooperative model of agent interaction. We introduce two communication protoc...
http://arxiv.org/pdf/1804.03980v1
2018-04-11T13:48:08
2018-04-11T13:48:08
{ "title": [ -0.012435913, 0.00932312, -0.0093688965, -0.021102905, -0.009094238, -0.0039596558, 0.0011730194, -0.022033691, -0.023208618, 0.023345947, -0.0057296753, -0.00008332729, -0.03616333, -0.03540039, -0.024963379, -0.012969971, -0.0016870499...
http://arxiv.org/abs/1804.03984v1
Emergence of Linguistic Communication from Referential Games with Symbolic and Pixel Input
Angeliki Lazaridou, Karl Moritz Hermann, Karl Tuyls, Stephen Clark
cs.AI, cs.CL, cs.LG, cs.MA
The ability of algorithms to evolve or learn (compositional) communication protocols has traditionally been studied in the language evolution literature through the use of emergent communication tasks. Here we scale up this research by using contemporary deep learning methods and by training reinforcement-learning neur...
http://arxiv.org/pdf/1804.03984v1
2018-04-11T13:51:19
2018-04-11T13:51:19
{ "title": [ -0.010177612, -0.0016517639, 0.01676941, 0.018798828, -0.02909851, 0.0037250519, 0.01802063, -0.04611206, -0.017852783, 0.02784729, 0.013130188, 0.00041985512, -0.014907837, -0.028869629, -0.043121338, -0.0132369995, -0.013931274, 0....
http://arxiv.org/abs/2004.04722v1
Re-conceptualising the Language Game Paradigm in the Framework of Multi-Agent Reinforcement Learning
Paul Van Eecke, Katrien Beuls
cs.AI, cs.CL, cs.LG, cs.MA
In this paper, we formulate the challenge of re-conceptualising the language game experimental paradigm in the framework of multi-agent reinforcement learning (MARL). If successful, future language game experiments will benefit from the rapid and promising methodological advances in the MARL community, while future MAR...
http://arxiv.org/pdf/2004.04722v1
2020-04-09T17:55:15
2020-04-09T17:55:15
{ "title": [ -0.027496338, -0.0024261475, -0.022628784, -0.005996704, -0.0007457733, -0.018859863, 0.014526367, -0.008483887, -0.040740967, 0.018997192, -0.0032291412, 0.010887146, -0.0501709, -0.02406311, -0.018371582, -0.029388428, -0.021942139, ...
http://arxiv.org/abs/2007.09820v1
Reinforcement Communication Learning in Different Social Network Structures
Marina Dubova, Arseny Moskvichev, Robert Goldstone
cs.AI, cs.CL, cs.LG, cs.MA
Social network structure is one of the key determinants of human language evolution. Previous work has shown that the network of social interactions shapes decentralized learning in human groups, leading to the emergence of different kinds of communicative conventions. We examined the effects of social network organiza...
http://arxiv.org/pdf/2007.09820v1
2020-07-19T23:57:30
2020-07-19T23:57:30
{ "title": [ -0.009803772, 0.014442444, -0.03604126, -0.011314392, 0.01802063, -0.024291992, 0.0056114197, -0.020126343, -0.04473877, 0.011581421, -0.005016327, 0.016830444, -0.039916992, -0.02319336, -0.00050878525, -0.018585205, -0.009773254, 0...
http://arxiv.org/abs/2304.07297v2
Language Instructed Reinforcement Learning for Human-AI Coordination
Hengyuan Hu, Dorsa Sadigh
cs.AI, cs.CL, cs.LG, cs.MA
One of the fundamental quests of AI is to produce agents that coordinate well with humans. This problem is challenging, especially in domains that lack high quality human behavioral data, because multi-agent reinforcement learning (RL) often converges to different equilibria from the ones that humans prefer. We propose...
http://arxiv.org/pdf/2304.07297v2
2023-06-10T20:00:45
2023-04-13T04:47:31
{ "title": [ -0.023925781, -0.0059165955, -0.028289795, -0.025314331, -0.0076065063, -0.020736694, -0.0035858154, -0.0014724731, -0.034820557, 0.03390503, -0.027694702, 0.018692017, -0.01612854, -0.0018520355, -0.018844604, -0.021560669, -0.007694244...
http://arxiv.org/abs/1905.12561v4
Anti-efficient encoding in emergent communication
Rahma Chaabouni, Eugene Kharitonov, Emmanuel Dupoux, Marco Baroni
cs.CL, cs.AI, cs.LG, cs.MA
Despite renewed interest in emergent language simulations with neural networks, little is known about the basic properties of the induced code, and how they compare to human language. One fundamental characteristic of the latter, known as Zipf's Law of Abbreviation (ZLA), is that more frequent words are efficiently ass...
http://arxiv.org/pdf/1905.12561v4
2019-10-15T11:56:14
2019-05-29T16:14:24
{ "title": [ -0.024215698, 0.0027275085, -0.017944336, -0.00060129166, -0.005657196, -0.0023822784, 0.023971558, -0.019805908, 0.019241333, 0.046966553, 0.0032157898, 0.0029773712, 0.024017334, -0.02432251, -0.020248413, 0.030090332, 0.0026016235, ...
http://arxiv.org/abs/2111.06345v1
Poisoning Knowledge Graph Embeddings via Relation Inference Patterns
Peru Bhardwaj, John Kelleher, Luca Costabello, Declan O'Sullivan
cs.LG, cs.AI, cs.CL, cs.NE
We study the problem of generating data poisoning attacks against Knowledge Graph Embedding (KGE) models for the task of link prediction in knowledge graphs. To poison KGE models, we propose to exploit their inductive abilities which are captured through the relationship patterns like symmetry, inversion and compositio...
http://arxiv.org/pdf/2111.06345v1
2021-11-11T17:57:37
2021-11-11T17:57:37
{ "title": [ 0.0017585754, -0.01701355, -0.046173096, -0.09234619, 0.0052375793, 0.001584053, -0.03677368, 0.019973755, -0.004180908, 0.003211975, -0.028762817, 0.012359619, -0.010055542, -0.01272583, -0.00881958, -0.013511658, -0.0053901672, 0.0...
http://arxiv.org/abs/2111.03120v1
Adversarial Attacks on Knowledge Graph Embeddings via Instance Attribution Methods
Peru Bhardwaj, John Kelleher, Luca Costabello, Declan O'Sullivan
cs.LG, cs.AI, cs.CL, cs.NE
Despite the widespread use of Knowledge Graph Embeddings (KGE), little is known about the security vulnerabilities that might disrupt their intended behaviour. We study data poisoning attacks against KGE models for link prediction. These attacks craft adversarial additions or deletions at training time to cause model f...
http://arxiv.org/pdf/2111.03120v1
2021-11-04T19:38:48
2021-11-04T19:38:48
{ "title": [ 0.0008749962, -0.015701294, -0.040283203, -0.03265381, 0.010253906, -0.02243042, -0.01525116, -0.021255493, -0.03475952, -0.0014047623, -0.0035362244, 0.017166138, -0.026016235, -0.034729004, 0.010787964, -0.021575928, 0.0011911392, ...
http://arxiv.org/abs/2202.10447v2
Transformer Quality in Linear Time
Weizhe Hua, Zihang Dai, Hanxiao Liu, Quoc V. Le
cs.LG, cs.AI, cs.CL, cs.NE
We revisit the design choices in Transformers, and propose methods to address their weaknesses in handling long sequences. First, we propose a simple layer named gated attention unit, which allows the use of a weaker single-head attention with minimal quality loss. We then propose a linear approximation method compleme...
http://arxiv.org/pdf/2202.10447v2
2022-06-27T17:59:19
2022-02-21T18:59:38
{ "title": [ 0.07019043, 0.019699097, 0.001953125, -0.014633179, -0.019104004, -0.0357666, 0.002948761, -0.041748047, -0.056610107, 0.043121338, -0.010803223, 0.01499176, 0.0005812645, -0.029663086, 0.028823853, 0.00035309792, 0.018035889, 0.0256...
http://arxiv.org/abs/2205.07006v1
Integration of Text and Graph-based Features for Detecting Mental Health Disorders from Voice
Nasser Ghadiri, Rasoul Samani, Fahime Shahrokh
cs.LG, cs.AI, cs.CL, cs.NE, I.2; J.3
With the availability of voice-enabled devices such as smart phones, mental health disorders could be detected and treated earlier, particularly post-pandemic. The current methods involve extracting features directly from audio signals. In this paper, two methods are used to enrich voice analysis for depression detecti...
http://arxiv.org/pdf/2205.07006v1
2022-05-14T08:37:19
2022-05-14T08:37:19
{ "title": [ 0.017745972, -0.00075626373, -0.040618896, -0.09375, 0.024276733, -0.002904892, -0.04888916, 0.017028809, 0.004535675, 0.03753662, -0.008163452, -0.008384705, 0.01612854, -0.034820557, -0.038238525, 0.0045814514, -0.014381409, 0.0524...
http://arxiv.org/abs/2209.04862v2
Adaptive Perturbation-Based Gradient Estimation for Discrete Latent Variable Models
Pasquale Minervini, Luca Franceschi, Mathias Niepert
cs.LG, cs.AI, cs.CL, cs.NE
The integration of discrete algorithmic components in deep learning architectures has numerous applications. Recently, Implicit Maximum Likelihood Estimation (IMLE, Niepert, Minervini, and Franceschi 2021), a class of gradient estimators for discrete exponential family distributions, was proposed by combining implicit ...
http://arxiv.org/pdf/2209.04862v2
2023-02-05T18:40:38
2022-09-11T13:32:39
{ "title": [ -0.028625488, 0.009124756, -0.047943115, -0.030029297, 0.013298035, -0.022033691, -0.028015137, -0.03262329, 0.00082206726, -0.008003235, -0.054779053, 0.010879517, -0.0357666, -0.030136108, 0.00020182133, -0.032989502, 0.018676758, ...
http://arxiv.org/abs/2205.10113v1
Evolutionary Multi-Armed Bandits with Genetic Thompson Sampling
Baihan Lin
cs.NE, cs.AI, cs.LG, cs.MA, stat.ML
As two popular schools of machine learning, online learning and evolutionary computations have become two important driving forces behind real-world decision making engines for applications in biomedicine, economics, and engineering fields. Although there are prior work that utilizes bandits to improve evolutionary alg...
http://arxiv.org/pdf/2205.10113v1
2022-04-26T22:41:17
2022-04-26T22:41:17
{ "title": [ -0.042419434, -0.0052490234, -0.0042800903, -0.020706177, 0.0024986267, 0.008277893, 0.018722534, -0.017318726, -0.01751709, 0.026123047, -0.04534912, 0.041046143, -0.052490234, -0.011360168, -0.035186768, -0.00075244904, -0.007167816, ...
http://arxiv.org/abs/2302.14838v3
EvoPrompting: Language Models for Code-Level Neural Architecture Search
Angelica Chen, David M. Dohan, David R. So
cs.NE, cs.AI, cs.CL, cs.LG
Given the recent impressive accomplishments of language models (LMs) for code generation, we explore the use of LMs as adaptive mutation and crossover operators for an evolutionary neural architecture search (NAS) algorithm. While NAS still proves too difficult a task for LMs to succeed at solely through prompting, we ...
http://arxiv.org/pdf/2302.14838v3
2023-11-16T18:02:19
2023-02-28T18:37:25
{ "title": [ -0.022567749, -0.019714355, -0.009925842, -0.03149414, -0.007423401, 0.0015287399, -0.022247314, -0.014289856, 0.0019950867, 0.031082153, -0.010772705, -0.003074646, -0.0053710938, -0.021759033, -0.052246094, 0.0068588257, -0.012214661, ...
http://arxiv.org/abs/1809.07098v1
Novelty-organizing team of classifiers in noisy and dynamic environments
Danilo Vasconcellos Vargas, Hirotaka Takano, Junichi Murata
cs.AI, cs.LG, cs.MA, cs.NE, cs.SY
In the real world, the environment is constantly changing with the input variables under the effect of noise. However, few algorithms were shown to be able to work under those circumstances. Here, Novelty-Organizing Team of Classifiers (NOTC) is applied to the continuous action mountain car as well as two variations of...
http://arxiv.org/pdf/1809.07098v1
2018-09-19T09:38:20
2018-09-19T09:38:20
{ "title": [ -0.008255005, 0.027313232, -0.03640747, -0.029891968, 0.032714844, -0.023864746, 0.016708374, -0.015281677, 0.0064926147, 0.0068893433, -0.041931152, 0.033599854, -0.057739258, 0.00001001358, -0.032592773, -0.025360107, 0.0011520386, ...
http://arxiv.org/abs/1804.02884v1
Policy Gradient With Value Function Approximation For Collective Multiagent Planning
Duc Thien Nguyen, Akshat Kumar, Hoong Chuin Lau
cs.AI, cs.LG, cs.MA, cs.NE, cs.SY
Decentralized (PO)MDPs provide an expressive framework for sequential decision making in a multiagent system. Given their computational complexity, recent research has focused on tractable yet practical subclasses of Dec-POMDPs. We address such a subclass called CDEC-POMDP where the collective behavior of a population ...
http://arxiv.org/pdf/1804.02884v1
2018-04-09T09:45:29
2018-04-09T09:45:29
{ "title": [ -0.004058838, 0.0068511963, -0.022872925, -0.04360962, 0.020889282, -0.005168915, 0.016311646, -0.006706238, -0.031402588, 0.014678955, -0.048706055, 0.033599854, -0.026779175, 0.012077332, -0.013023376, -0.032073975, -0.014030457, 0...
http://arxiv.org/abs/1708.02977v1
Hierarchically-Attentive RNN for Album Summarization and Storytelling
Licheng Yu, Mohit Bansal, Tamara L. Berg
cs.CL, cs.AI, cs.CV, cs.LG
We address the problem of end-to-end visual storytelling. Given a photo album, our model first selects the most representative (summary) photos, and then composes a natural language story for the album. For this task, we make use of the Visual Storytelling dataset and a model composed of three hierarchically-attentive ...
http://arxiv.org/pdf/1708.02977v1
2017-08-09T19:26:47
2017-08-09T19:26:47
{ "title": [ -0.026794434, -0.0209198, -0.04751587, -0.07122803, -0.028823853, 0.0025901794, -0.06088257, 0.03274536, -0.0289917, 0.044158936, -0.018371582, 0.01626587, 0.022064209, -0.004814148, -0.024108887, 0.0016202927, -0.013908386, 0.025451...
http://arxiv.org/abs/1808.10009v1
Learning a Policy for Opportunistic Active Learning
Aishwarya Padmakumar, Peter Stone, Raymond J. Mooney
cs.CL, cs.AI, cs.CV, cs.LG
Active learning identifies data points to label that are expected to be the most useful in improving a supervised model. Opportunistic active learning incorporates active learning into interactive tasks that constrain possible queries during interactions. Prior work has shown that opportunistic active learning can be u...
http://arxiv.org/pdf/1808.10009v1
2018-08-29T18:40:26
2018-08-29T18:40:26
{ "title": [ 0.010955811, 0.0011196136, -0.009788513, -0.049743652, 0.017211914, -0.006641388, 0.0055351257, -0.020462036, -0.023208618, 0.012207031, -0.031677246, 0.03967285, -0.06341553, -0.015930176, -0.0079956055, -0.0076942444, -0.0027236938, ...
http://arxiv.org/abs/1706.01322v2
Deep learning evaluation using deep linguistic processing
Alexander Kuhnle, Ann Copestake
cs.CL, cs.AI, cs.CV, cs.LG
We discuss problems with the standard approaches to evaluation for tasks like visual question answering, and argue that artificial data can be used to address these as a complement to current practice. We demonstrate that with the help of existing 'deep' linguistic processing technology we are able to create challengin...
http://arxiv.org/pdf/1706.01322v2
2018-05-12T10:37:02
2017-06-05T13:53:56
{ "title": [ -0.025344849, 0.006843567, -0.03945923, -0.041870117, 0.020736694, -0.0131073, -0.014572144, -0.021713257, -0.005771637, 0.020767212, -0.020324707, -0.007896423, -0.016677856, -0.02017212, -0.049194336, -0.005470276, -0.017028809, 0....
http://arxiv.org/abs/1806.00738v1
Contextualize, Show and Tell: A Neural Visual Storyteller
Diana Gonzalez-Rico, Gibran Fuentes-Pineda
cs.CL, cs.AI, cs.CV, cs.LG
We present a neural model for generating short stories from image sequences, which extends the image description model by Vinyals et al. (Vinyals et al., 2015). This extension relies on an encoder LSTM to compute a context vector of each story from the image sequence. This context vector is used as the first state of m...
http://arxiv.org/pdf/1806.00738v1
2018-06-03T05:09:54
2018-06-03T05:09:54
{ "title": [ -0.017868042, -0.00014221668, -0.039398193, -0.04522705, -0.0011796951, -0.009117126, -0.020690918, -0.011543274, -0.013923645, 0.038970947, -0.00749588, -0.007331848, -0.005744934, -0.015808105, -0.047180176, -0.007904053, -0.0074882507...
http://arxiv.org/abs/2007.06898v1
Our Evaluation Metric Needs an Update to Encourage Generalization
Swaroop Mishra, Anjana Arunkumar, Chris Bryan, Chitta Baral
cs.CL, cs.AI, cs.CV, cs.LG
Models that surpass human performance on several popular benchmarks display significant degradation in performance on exposure to Out of Distribution (OOD) data. Recent research has shown that models overfit to spurious biases and `hack' datasets, in lieu of learning generalizable features like humans. In order to stop...
http://arxiv.org/pdf/2007.06898v1
2020-07-14T08:15:19
2020-07-14T08:15:19
{ "title": [ 0.01499176, 0.008087158, -0.0006117821, -0.015930176, -0.018234253, 0.019454956, 0.02029419, 0.0104904175, -0.024719238, 0.032104492, -0.019195557, 0.0006737709, -0.054901123, -0.032043457, 0.0073013306, -0.0231781, 0.028915405, 0.00...
http://arxiv.org/abs/2305.10865v2
Semantically Aligned Task Decomposition in Multi-Agent Reinforcement Learning
Wenhao Li, Dan Qiao, Baoxiang Wang, Xiangfeng Wang, Bo Jin, Hongyuan Zha
cs.LG, cs.AI, cs.CL, cs.MA
The difficulty of appropriately assigning credit is particularly heightened in cooperative MARL with sparse reward, due to the concurrent time and structural scales involved. Automatic subgoal generation (ASG) has recently emerged as a viable MARL approach inspired by utilizing subgoals in intrinsically motivated reinf...
http://arxiv.org/pdf/2305.10865v2
2023-09-30T08:27:28
2023-05-18T10:37:54
{ "title": [ -0.019302368, 0.016601562, -0.04714966, -0.011787415, 0.02229309, -0.020553589, 0.0068855286, -0.014457703, -0.034179688, 0.020385742, -0.006389618, 0.031951904, -0.043151855, -0.008514404, -0.008308411, -0.01939392, -0.007888794, 0....
http://arxiv.org/abs/2212.07891v1
Emergent Behaviors in Multi-Agent Target Acquisition
Piyush K. Sharma, Erin Zaroukian, Derrik E. Asher, Bryson Howell
cs.AI, cs.CV, cs.LG, cs.MA
Only limited studies and superficial evaluations are available on agents' behaviors and roles within a Multi-Agent System (MAS). We simulate a MAS using Reinforcement Learning (RL) in a pursuit-evasion (a.k.a predator-prey pursuit) game, which shares task goals with target acquisition, and we create different adversari...
http://arxiv.org/pdf/2212.07891v1
2022-12-15T15:20:58
2022-12-15T15:20:58
{ "title": [ -0.0063552856, 0.022857666, -0.01399231, -0.014518738, 0.022460938, -0.015136719, 0.035888672, -0.023529053, -0.03338623, 0.035949707, -0.009414673, 0.034698486, -0.049682617, -0.0013685226, -0.030090332, -0.00605011, -0.009628296, 0...
http://arxiv.org/abs/1909.10476v1
Research Directions in Democratizing Innovation through Design Automation, One-Click Manufacturing Services and Intelligent Machines
Binil Starly, Atin Angrish, Paul Cohen
cs.CY, cs.AI, cs.CV, cs.LG, cs.MA
The digitalization of manufacturing has created opportunities for consumers to customize products that fit their individualized needs which in turn would drive demand for manufacturing services. However, this pull-based manufacturing system production of extremely low quantity and limitless variety for products is expe...
http://arxiv.org/pdf/1909.10476v1
2019-09-23T16:56:08
2019-09-23T16:56:08
{ "title": [ 0.0020961761, 0.004737854, -0.07507324, -0.00774765, 0.011756897, -0.058380127, 0.020721436, 0.013839722, -0.029891968, 0.04248047, -0.0042266846, 0.037994385, 0.001162529, -0.02027893, -0.0050811768, -0.01158905, -0.007659912, 0.030...
http://arxiv.org/abs/1102.2739v1
A General Framework for Development of the Cortex-like Visual Object Recognition System: Waves of Spikes, Predictive Coding and Universal Dictionary of Features
Sergey S. Tarasenko
cs.CV, cs.AI, cs.LG, cs.NE
This study is focused on the development of the cortex-like visual object recognition system. We propose a general framework, which consists of three hierarchical levels (modules). These modules functionally correspond to the V1, V4 and IT areas. Both bottom-up and top-down connections between the hierarchical levels V...
http://arxiv.org/pdf/1102.2739v1
2011-02-14T11:40:08
2011-02-14T11:40:08
{ "title": [ -0.006385803, -0.0026054382, -0.03768921, -0.033050537, -0.020874023, -0.010284424, -0.02267456, 0.0072364807, 0.01474762, 0.01586914, -0.03781128, 0.00504303, -0.011398315, -0.029083252, -0.02444458, -0.010528564, 0.029724121, 0.020...
http://arxiv.org/abs/1512.04295v2
Origami: A 803 GOp/s/W Convolutional Network Accelerator
Lukas Cavigelli, Luca Benini
cs.CV, cs.AI, cs.LG, cs.NE, B.7.1; I.2.6
An ever increasing number of computer vision and image/video processing challenges are being approached using deep convolutional neural networks, obtaining state-of-the-art results in object recognition and detection, semantic segmentation, action recognition, optical flow and superresolution. Hardware acceleration of ...
http://arxiv.org/pdf/1512.04295v2
2016-01-19T22:56:41
2015-12-14T13:06:43
{ "title": [ -0.06530762, -0.020431519, -0.05911255, -0.035064697, -0.0013246536, -0.0014448166, -0.03363037, 0.006664276, -0.029464722, 0.027130127, -0.025375366, 0.011398315, -0.0019054413, -0.023834229, -0.016204834, -0.008781433, 0.006099701, ...
http://arxiv.org/abs/1605.06431v2
Residual Networks Behave Like Ensembles of Relatively Shallow Networks
Andreas Veit, Michael Wilber, Serge Belongie
cs.CV, cs.AI, cs.LG, cs.NE
In this work we propose a novel interpretation of residual networks showing that they can be seen as a collection of many paths of differing length. Moreover, residual networks seem to enable very deep networks by leveraging only the short paths during training. To support this observation, we rewrite residual networks...
http://arxiv.org/pdf/1605.06431v2
2016-10-27T00:43:58
2016-05-20T16:44:03
{ "title": [ 0.0155181885, -0.0010051727, -0.0076026917, -0.00042271614, -0.01499939, -0.023544312, 0.0011768341, -0.019714355, 0.010375977, 0.0022506714, -0.04559326, -0.01158905, -0.0059280396, -0.029525757, -0.0018577576, -0.024032593, 0.003530502...
http://arxiv.org/abs/1607.01719v1
Deep CORAL: Correlation Alignment for Deep Domain Adaptation
Baochen Sun, Kate Saenko
cs.CV, cs.AI, cs.LG, cs.NE
Deep neural networks are able to learn powerful representations from large quantities of labeled input data, however they cannot always generalize well across changes in input distributions. Domain adaptation algorithms have been proposed to compensate for the degradation in performance due to domain shift. In this pap...
http://arxiv.org/pdf/1607.01719v1
2016-07-06T17:35:55
2016-07-06T17:35:55
{ "title": [ -0.020690918, 0.013557434, -0.05886841, -0.04537964, 0.018875122, -0.02659607, -0.019104004, -0.005657196, 0.015510559, 0.043914795, 0.002380371, 0.0038223267, 0.015327454, -0.015525818, -0.04434204, 0.023513794, -0.0030956268, 0.023...
http://arxiv.org/abs/1607.07043v1
Spatio-Temporal LSTM with Trust Gates for 3D Human Action Recognition
Jun Liu, Amir Shahroudy, Dong Xu, Gang Wang
cs.CV, cs.AI, cs.LG, cs.NE
3D action recognition - analysis of human actions based on 3D skeleton data - becomes popular recently due to its succinctness, robustness, and view-invariant representation. Recent attempts on this problem suggested to develop RNN-based learning methods to model the contextual dependency in the temporal domain. In thi...
http://arxiv.org/pdf/1607.07043v1
2016-07-24T13:39:11
2016-07-24T13:39:11
{ "title": [ 0.013069153, 0.0335083, -0.06915283, -0.054718018, 0.027740479, -0.02947998, -0.014831543, -0.009048462, -0.020492554, 0.045928955, -0.044128418, 0.006752014, 0.025375366, -0.010147095, -0.034362793, -0.02394104, 0.0067977905, 0.0063...
http://arxiv.org/abs/1612.01294v1
Message Passing Multi-Agent GANs
Arnab Ghosh, Viveka Kulharia, Vinay Namboodiri
cs.CV, cs.AI, cs.LG, cs.NE
Communicating and sharing intelligence among agents is an important facet of achieving Artificial General Intelligence. As a first step towards this challenge, we introduce a novel framework for image generation: Message Passing Multi-Agent Generative Adversarial Networks (MPM GANs). While GANs have recently been shown...
http://arxiv.org/pdf/1612.01294v1
2016-12-05T10:10:13
2016-12-05T10:10:13
{ "title": [ -0.03186035, 0.01828003, -0.03866577, -0.03366089, 0.025863647, -0.011672974, -0.008636475, -0.00699234, 0.0028858185, 0.03189087, -0.011741638, 0.006969452, 0.0006966591, -0.01600647, -0.025726318, 0.0053863525, -0.015396118, 0.0024...
http://arxiv.org/abs/1707.01357v1
Improving Content-Invariance in Gated Autoencoders for 2D and 3D Object Rotation
Stefan Lattner, Maarten Grachten
cs.CV, cs.AI, cs.LG, cs.NE
Content-invariance in mapping codes learned by GAEs is a useful feature for various relation learning tasks. In this paper we show that the content-invariance of mapping codes for images of 2D and 3D rotated objects can be substantially improved by extending the standard GAE loss (symmetric reconstruction error) with a...
http://arxiv.org/pdf/1707.01357v1
2017-07-05T12:28:43
2017-07-05T12:28:43
{ "title": [ -0.01852417, 0.039154053, -0.06616211, -0.008720398, 0.022445679, -0.03643799, -0.012023926, -0.028900146, 0.008651733, 0.05670166, -0.04437256, 0.026672363, 0.022415161, -0.0018529892, -0.03543091, 0.02218628, 0.010795593, 0.0152664...
http://arxiv.org/abs/1707.03502v2
Deep Learning for Sensor-based Activity Recognition: A Survey
Jindong Wang, Yiqiang Chen, Shuji Hao, Xiaohui Peng, Lisha Hu
cs.CV, cs.AI, cs.LG, cs.NE
Sensor-based activity recognition seeks the profound high-level knowledge about human activities from multitudes of low-level sensor readings. Conventional pattern recognition approaches have made tremendous progress in the past years. However, those methods often heavily rely on heuristic hand-crafted feature extracti...
http://arxiv.org/pdf/1707.03502v2
2017-12-14T03:11:15
2017-07-12T00:21:04
{ "title": [ -0.011474609, 0.026504517, -0.04537964, -0.038238525, 0.033813477, -0.022506714, -0.016738892, -0.024795532, -0.011680603, 0.024139404, -0.036834717, 0.016983032, -0.025482178, -0.013298035, -0.03466797, -0.010368347, -0.0116119385, ...
http://arxiv.org/abs/1904.03816v1
Towards Real-Time Automatic Portrait Matting on Mobile Devices
Seokjun Seo, Seungwoo Choi, Martin Kersner, Beomjun Shin, Hyungsuk Yoon, Hyeongmin Byun, Sungjoo Ha
cs.CV, cs.AI, cs.LG, cs.NE
We tackle the problem of automatic portrait matting on mobile devices. The proposed model is aimed at attaining real-time inference on mobile devices with minimal degradation of model performance. Our model MMNet, based on multi-branch dilated convolution with linear bottleneck blocks, outperforms the state-of-the-art ...
http://arxiv.org/pdf/1904.03816v1
2019-04-08T03:21:25
2019-04-08T03:21:25
{ "title": [ -0.030593872, 0.004878998, -0.02558899, -0.045776367, 0.014961243, -0.0017414093, -0.029769897, -0.0046195984, -0.03378296, 0.053344727, 0.0029830933, 0.043914795, 0.010269165, -0.013244629, -0.02670288, 0.0184021, 0.007255554, -0.00...
http://arxiv.org/abs/1906.01558v2
Disentangling neural mechanisms for perceptual grouping
Junkyung Kim, Drew Linsley, Kalpit Thakkar, Thomas Serre
cs.CV, cs.AI, cs.LG, cs.NE
Forming perceptual groups and individuating objects in visual scenes is an essential step towards visual intelligence. This ability is thought to arise in the brain from computations implemented by bottom-up, horizontal, and top-down connections between neurons. However, the relative contributions of these connections ...
http://arxiv.org/pdf/1906.01558v2
2020-10-28T15:53:55
2019-06-04T16:21:46
{ "title": [ 0.0002963543, 0.008491516, -0.05722046, -0.057678223, 0.027908325, -0.026321411, 0.011108398, -0.0064926147, -0.0050811768, 0.04296875, -0.01159668, -0.0051231384, -0.01386261, -0.011398315, -0.051513672, 0.0030136108, 0.007347107, 0...
http://arxiv.org/abs/1903.09876v1
Automated pulmonary nodule detection using 3D deep convolutional neural networks
Hao Tang, Daniel R. Kim, Xiaohui Xie
cs.CV, cs.AI, cs.LG, cs.NE
Early detection of pulmonary nodules in computed tomography (CT) images is essential for successful outcomes among lung cancer patients. Much attention has been given to deep convolutional neural network (DCNN)-based approaches to this task, but models have relied at least partly on 2D or 2.5D components for inherently...
http://arxiv.org/pdf/1903.09876v1
2019-03-23T20:20:15
2019-03-23T20:20:15
{ "title": [ 0.0051193237, 0.012069702, -0.02973938, -0.047210693, 0.017333984, -0.0046920776, -0.021530151, -0.01637268, -0.005432129, 0.032287598, -0.05166626, -0.00016498566, -0.0051994324, -0.008346558, -0.05142212, -0.01524353, 0.029815674, ...
http://arxiv.org/abs/1511.05547v2
Return of Frustratingly Easy Domain Adaptation
Baochen Sun, Jiashi Feng, Kate Saenko
cs.CV, cs.AI, cs.LG, cs.NE
Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of conventional machine learning methods. Supervised domain adaptation methods have been proposed for ...
http://arxiv.org/pdf/1511.05547v2
2015-12-09T05:39:43
2015-11-17T20:53:26
{ "title": [ 0.0146865845, -0.02142334, -0.017852783, -0.015930176, -0.046722412, -0.03289795, 0.014579773, 0.01965332, 0.020858765, 0.035858154, -0.021438599, -0.009422302, 0.027542114, -0.038085938, -0.0032596588, 0.006046295, -0.0064697266, 0....
http://arxiv.org/abs/1805.08191v3
Hierarchically Structured Reinforcement Learning for Topically Coherent Visual Story Generation
Qiuyuan Huang, Zhe Gan, Asli Celikyilmaz, Dapeng Wu, Jianfeng Wang, Xiaodong He
cs.CV, cs.AI, cs.LG, cs.NE
We propose a hierarchically structured reinforcement learning approach to address the challenges of planning for generating coherent multi-sentence stories for the visual storytelling task. Within our framework, the task of generating a story given a sequence of images is divided across a two-level hierarchical decoder...
http://arxiv.org/pdf/1805.08191v3
2019-01-18T07:58:43
2018-05-21T17:23:31
{ "title": [ -0.013252258, -0.004798889, -0.040924072, -0.061401367, -0.009979248, -0.0012617111, -0.031204224, -0.0015640259, -0.021026611, 0.027053833, -0.01689148, 0.024841309, -0.025161743, -0.010116577, -0.027694702, -0.021453857, -0.015914917, ...
http://arxiv.org/abs/1805.10795v1
Deep Discriminative Latent Space for Clustering
Elad Tzoreff, Olga Kogan, Yoni Choukroun
cs.CV, cs.AI, cs.LG, cs.NE
Clustering is one of the most fundamental tasks in data analysis and machine learning. It is central to many data-driven applications that aim to separate the data into groups with similar patterns. Moreover, clustering is a complex procedure that is affected significantly by the choice of the data representation metho...
http://arxiv.org/pdf/1805.10795v1
2018-05-28T07:34:14
2018-05-28T07:34:14
{ "title": [ -0.0025253296, 0.007980347, -0.059539795, -0.05618286, 0.036987305, -0.010665894, -0.017425537, -0.020492554, 0.009475708, 0.017364502, -0.022384644, 0.022232056, -0.024673462, -0.020111084, -0.018829346, 0.0072517395, 0.008140564, 0...
http://arxiv.org/abs/1811.02234v1
Semantic bottleneck for computer vision tasks
Maxime Bucher, Stéphane Herbin, Frédéric Jurie
cs.CV, cs.AI, cs.LG, cs.NE
This paper introduces a novel method for the representation of images that is semantic by nature, addressing the question of computation intelligibility in computer vision tasks. More specifically, our proposition is to introduce what we call a semantic bottleneck in the processing pipeline, which is a crossing point i...
http://arxiv.org/pdf/1811.02234v1
2018-11-06T09:01:02
2018-11-06T09:01:02
{ "title": [ -0.030273438, -0.005718231, -0.039215088, -0.039245605, 0.004196167, -0.013771057, -0.018447876, -0.0026302338, 0.019378662, 0.029449463, -0.054107666, 0.008338928, -0.013771057, 0.0014152527, -0.033203125, 0.0002989769, 0.022613525, ...
http://arxiv.org/abs/2011.12930v2
Unsupervised Object Keypoint Learning using Local Spatial Predictability
Anand Gopalakrishnan, Sjoerd van Steenkiste, Jürgen Schmidhuber
cs.CV, cs.AI, cs.LG, cs.NE
We propose PermaKey, a novel approach to representation learning based on object keypoints. It leverages the predictability of local image regions from spatial neighborhoods to identify salient regions that correspond to object parts, which are then converted to keypoints. Unlike prior approaches, it utilizes predictab...
http://arxiv.org/pdf/2011.12930v2
2021-03-08T15:10:29
2020-11-25T18:27:05
{ "title": [ -0.014259338, 0.02507019, -0.02041626, -0.041778564, 0.022720337, 0.012542725, -0.020111084, -0.049682617, 0.028396606, 0.0037765503, -0.026672363, 0.005821228, -0.032226562, -0.0038394928, -0.039276123, -0.0011329651, 0.015823364, 0...
http://arxiv.org/abs/2102.05451v1
Two Novel Performance Improvements for Evolving CNN Topologies
Yaron Strauch, Jo Grundy
cs.CV, cs.AI, cs.LG, cs.NE, I.2.6; I.2.8; I.2.10; I.5.1; I.5.2; I.6.m
Convolutional Neural Networks (CNNs) are the state-of-the-art algorithms for the processing of images. However the configuration and training of these networks is a complex task requiring deep domain knowledge, experience and much trial and error. Using genetic algorithms, competitive CNN topologies for image recogniti...
http://arxiv.org/pdf/2102.05451v1
2021-02-10T14:17:51
2021-02-10T14:17:51
{ "title": [ -0.029891968, 0.008903503, -0.024932861, -0.02192688, 0.024246216, -0.006290436, 0.0047073364, -0.0054397583, -0.002166748, 0.043823242, -0.023101807, 0.013053894, -0.016357422, 0.0006213188, -0.03515625, 0.0107803345, 0.012580872, 0...
http://arxiv.org/abs/2103.03060v1
BM3D vs 2-Layer ONN
Junaid Malik, Serkan Kiranyaz, Mehmet Yamac, Moncef Gabbouj
cs.CV, cs.AI, cs.LG, cs.NE
Despite their recent success on image denoising, the need for deep and complex architectures still hinders the practical usage of CNNs. Older but computationally more efficient methods such as BM3D remain a popular choice, especially in resource-constrained scenarios. In this study, we aim to find out whether compact n...
http://arxiv.org/pdf/2103.03060v1
2021-03-04T14:37:23
2021-03-04T14:37:23
{ "title": [ 0.008483887, 0.00983429, -0.051971436, -0.020706177, -0.03656006, -0.025375366, -0.03717041, -0.010864258, -0.06488037, 0.023956299, 0.011245728, -0.017593384, 0.00756073, -0.049316406, 0.033203125, -0.024642944, 0.013496399, 0.02400...
http://arxiv.org/abs/2106.02328v1
Temporally coherent video anonymization through GAN inpainting
Thangapavithraa Balaji, Patrick Blies, Georg Göri, Raphael Mitsch, Marcel Wasserer, Torsten Schön
cs.CV, cs.AI, cs.LG, cs.NE
This work tackles the problem of temporally coherent face anonymization in natural video streams.We propose JaGAN, a two-stage system starting with detecting and masking out faces with black image patches in all individual frames of the video. The second stage leverages a privacy-preserving Video Generative Adversarial...
http://arxiv.org/pdf/2106.02328v1
2021-06-04T08:19:44
2021-06-04T08:19:44
{ "title": [ 0.0059013367, 0.029586792, -0.049804688, -0.04446411, 0.042297363, -0.009643555, 0.018371582, -0.00058317184, -0.028060913, 0.036895752, 0.0029029846, 0.01586914, 0.0049476624, -0.027542114, -0.026687622, -0.013771057, -0.02168274, 0...
http://arxiv.org/abs/2106.10944v2
Hard hat wearing detection based on head keypoint localization
Bartosz Wójcik, Mateusz Żarski, Kamil Książek, Jarosław Adam Miszczak, Mirosław Jan Skibniewski
cs.CV, cs.AI, cs.LG, cs.NE
In recent years, a lot of attention is paid to deep learning methods in the context of vision-based construction site safety systems, especially regarding personal protective equipment. However, despite all this attention, there is still no reliable way to establish the relationship between workers and their hard hats....
http://arxiv.org/pdf/2106.10944v2
2022-06-24T10:27:20
2021-06-21T09:31:33
{ "title": [ -0.029296875, 0.012184143, -0.031311035, -0.0077934265, -0.003030777, -0.017303467, -0.0037460327, -0.026229858, -0.010124207, 0.036956787, -0.03555298, 0.044189453, 0.007881165, -0.011993408, -0.01727295, 0.019973755, 0.009552002, 0...
http://arxiv.org/abs/2202.13072v1
Adversarial Contrastive Self-Supervised Learning
Wentao Zhu, Hang Shang, Tingxun Lv, Chao Liao, Sen Yang, Ji Liu
cs.CV, cs.AI, cs.LG, cs.NE
Recently, learning from vast unlabeled data, especially self-supervised learning, has been emerging and attracted widespread attention. Self-supervised learning followed by the supervised fine-tuning on a few labeled examples can significantly improve label efficiency and outperform standard supervised training using f...
http://arxiv.org/pdf/2202.13072v1
2022-02-26T05:57:45
2022-02-26T05:57:45
{ "title": [ -0.004322052, 0.019378662, -0.025161743, -0.039794922, 0.027114868, -0.007194519, -0.010055542, -0.03942871, -0.018127441, 0.02229309, -0.0035552979, 0.022613525, -0.04525757, -0.026565552, -0.032562256, -0.0011301041, 0.001950264, 0...
http://arxiv.org/abs/2204.06214v1
Context-based Deep Learning Architecture with Optimal Integration Layer for Image Parsing
Ranju Mandal, Basim Azam, Brijesh Verma
cs.CV, cs.AI, cs.LG, cs.NE
Deep learning models have been efficient lately on image parsing tasks. However, deep learning models are not fully capable of exploiting visual and contextual information simultaneously. The proposed three-layer context-based deep architecture is capable of integrating context explicitly with visual information. The n...
http://arxiv.org/pdf/2204.06214v1
2022-04-13T07:35:39
2022-04-13T07:35:39
{ "title": [ 0.003944397, 0.009391785, -0.032196045, -0.06768799, 0.029129028, -0.007457733, -0.0146865845, -0.004108429, -0.0053367615, 0.041412354, -0.026153564, 0.012214661, -0.010292053, -0.010025024, -0.035736084, -0.0008983612, 0.0056037903, ...
http://arxiv.org/abs/2204.11887v2
Evolutionary latent space search for driving human portrait generation
Benjamín Machín, Sergio Nesmachnow, Jamal Toutouh
cs.CV, cs.AI, cs.LG, cs.NE
This article presents an evolutionary approach for synthetic human portraits generation based on the latent space exploration of a generative adversarial network. The idea is to produce different human face images very similar to a given target portrait. The approach applies StyleGAN2 for portrait generation and FaceNe...
http://arxiv.org/pdf/2204.11887v2
2022-05-19T13:16:23
2022-04-25T18:00:49
{ "title": [ -0.007621765, 0.00026726723, -0.0038795471, -0.024230957, 0.002073288, 0.0026988983, 0.007888794, -0.028259277, 0.0033473969, 0.038208008, -0.012840271, 0.034423828, -0.002248764, -0.015075684, -0.0259552, 0.02166748, 0.012306213, -0...
http://arxiv.org/abs/2205.05167v2
Robustness of Humans and Machines on Object Recognition with Extreme Image Transformations
Dakarai Crowder, Girik Malik
cs.CV, cs.AI, cs.LG, cs.NE
Recent neural network architectures have claimed to explain data from the human visual cortex. Their demonstrated performance is however still limited by the dependence on exploiting low-level features for solving visual tasks. This strategy limits their performance in case of out-of-distribution/adversarial data. Huma...
http://arxiv.org/pdf/2205.05167v2
2022-05-27T05:04:35
2022-05-09T17:15:54
{ "title": [ 0.01109314, 0.04067993, -0.03326416, -0.0031547546, 0.04598999, -0.037078857, 0.030731201, -0.04269409, 0.007873535, 0.02178955, -0.04135132, 0.013954163, -0.015838623, -0.0036449432, -0.057769775, -0.009468079, 0.015258789, 0.022521...
http://arxiv.org/abs/2205.08836v2
Large Neural Networks Learning from Scratch with Very Few Data and without Explicit Regularization
Christoph Linse, Thomas Martinetz
cs.CV, cs.AI, cs.LG, cs.NE
Recent findings have shown that highly over-parameterized Neural Networks generalize without pretraining or explicit regularization. It is achieved with zero training error, i.e., complete over-fitting by memorizing the training data. This is surprising, since it is completely against traditional machine learning wisdo...
http://arxiv.org/pdf/2205.08836v2
2022-10-21T07:37:37
2022-05-18T10:08:28
{ "title": [ -0.0020065308, 0.002149582, -0.046966553, -0.05142212, 0.027328491, -0.03201294, -0.00040221214, -0.007675171, 0.0011911392, 0.0146484375, -0.028717041, 0.010696411, -0.027053833, -0.027740479, -0.015159607, -0.004219055, 0.0209198, ...
http://arxiv.org/abs/2209.04362v2
EDeNN: Event Decay Neural Networks for low latency vision
Celyn Walters, Simon Hadfield
cs.CV, cs.AI, cs.LG, cs.NE
Despite the success of neural networks in computer vision tasks, digital 'neurons' are a very loose approximation of biological neurons. Today's learning approaches are designed to function on digital devices with digital data representations such as image frames. In contrast, biological vision systems are generally mu...
http://arxiv.org/pdf/2209.04362v2
2023-05-09T14:22:17
2022-09-09T15:51:39
{ "title": [ 0.0028705597, -0.0073280334, -0.026733398, -0.05593872, 0.030044556, -0.010620117, -0.004550934, -0.020339966, 0.01689148, 0.049957275, -0.0032844543, -0.01373291, 0.013679504, -0.013542175, -0.053833008, 0.012611389, 0.0032787323, 0...
http://arxiv.org/abs/2301.00620v1
Dynamically Modular and Sparse General Continual Learning
Arnav Varma, Elahe Arani, Bahram Zonooz
cs.CV, cs.AI, cs.LG, cs.NE
Real-world applications often require learning continuously from a stream of data under ever-changing conditions. When trying to learn from such non-stationary data, deep neural networks (DNNs) undergo catastrophic forgetting of previously learned information. Among the common approaches to avoid catastrophic forgettin...
http://arxiv.org/pdf/2301.00620v1
2023-01-02T12:24:24
2023-01-02T12:24:24
{ "title": [ -0.01398468, 0.0041160583, -0.07525635, -0.07678223, 0.036010742, -0.02407837, -0.022964478, 0.027374268, -0.0029659271, 0.031707764, -0.006778717, 0.015335083, -0.0234375, -0.013900757, -0.047912598, 0.012939453, 0.020309448, 0.0047...
http://arxiv.org/abs/2303.02034v2
Linear CNNs Discover the Statistical Structure of the Dataset Using Only the Most Dominant Frequencies
Hannah Pinson, Joeri Lenaerts, Vincent Ginis
cs.CV, cs.AI, cs.LG, cs.NE
We here present a stepping stone towards a deeper understanding of convolutional neural networks (CNNs) in the form of a theory of learning in linear CNNs. Through analyzing the gradient descent equations, we discover that the evolution of the network during training is determined by the interplay between the dataset s...
http://arxiv.org/pdf/2303.02034v2
2023-07-25T00:19:04
2023-03-03T15:52:06
{ "title": [ -0.008674622, 0.008430481, -0.038238525, 0.012840271, -0.02079773, -0.037261963, -0.0013666153, -0.0121536255, -0.027145386, 0.035095215, -0.01876831, 0.01058197, -0.014930725, -0.038146973, 0.0048332214, -0.02017212, 0.019485474, -0...
http://arxiv.org/abs/2310.17427v1
Handshape recognition for Argentinian Sign Language using ProbSom
Franco Ronchetti, Facundo Manuel Quiroga, César Estrebou, Laura Lanzarini
cs.CV, cs.AI, cs.LG, cs.NE
Automatic sign language recognition is an important topic within the areas of human-computer interaction and machine learning. On the one hand, it poses a complex challenge that requires the intervention of various knowledge areas, such as video processing, image processing, intelligent systems and linguistics. On the ...
http://arxiv.org/pdf/2310.17427v1
2023-10-26T14:32:44
2023-10-26T14:32:44
{ "title": [ -0.021438599, 0.023223877, -0.04977417, -0.020217896, -0.019119263, -0.003622055, -0.0541687, -0.023956299, 0.017166138, 0.03793335, -0.0008111, -0.030639648, 0.031921387, -0.045928955, -0.034729004, -0.0018997192, -0.009246826, 0.00...
http://arxiv.org/abs/2311.01617v1
Look-Ahead Selective Plasticity for Continual Learning of Visual Tasks
Rouzbeh Meshkinnejad, Jie Mei, Daniel Lizotte, Yalda Mohsenzadeh
cs.CV, cs.AI, cs.LG, cs.NE
Contrastive representation learning has emerged as a promising technique for continual learning as it can learn representations that are robust to catastrophic forgetting and generalize well to unseen future tasks. Previous work in continual learning has addressed forgetting by using previous task data and trained mode...
http://arxiv.org/pdf/2311.01617v1
2023-11-02T22:00:23
2023-11-02T22:00:23
{ "title": [ 0.006340027, 0.0041656494, -0.033599854, -0.046936035, 0.0079422, -0.030197144, 0.0040130615, -0.008682251, -0.0010480881, 0.017959595, -0.011657715, 0.0138168335, -0.02583313, -0.0206604, -0.05239868, 0.005420685, 0.0068473816, 0.01...
http://arxiv.org/abs/1608.02717v1
Mean Box Pooling: A Rich Image Representation and Output Embedding for the Visual Madlibs Task
Ashkan Mokarian, Mateusz Malinowski, Mario Fritz
cs.CV, cs.AI, cs.CL, cs.LG
We present Mean Box Pooling, a novel visual representation that pools over CNN representations of a large number, highly overlapping object proposals. We show that such representation together with nCCA, a successful multimodal embedding technique, achieves state-of-the-art performance on the Visual Madlibs task. Moreo...
http://arxiv.org/pdf/1608.02717v1
2016-08-09T08:24:02
2016-08-09T08:24:02
{ "title": [ 0.012886047, -0.0067596436, -0.036834717, -0.05923462, -0.002368927, 0.005203247, -0.031280518, -0.0008649826, 0.03302002, 0.042053223, 0.012214661, 0.013465881, -0.015609741, -0.024459839, -0.08111572, 0.021591187, 0.00907135, -0.00...
http://arxiv.org/abs/1811.05013v1
Blindfold Baselines for Embodied QA
Ankesh Anand, Eugene Belilovsky, Kyle Kastner, Hugo Larochelle, Aaron Courville
cs.CV, cs.AI, cs.CL, cs.LG
We explore blindfold (question-only) baselines for Embodied Question Answering. The EmbodiedQA task requires an agent to answer a question by intelligently navigating in a simulated environment, gathering necessary visual information only through first-person vision before finally answering. Consequently, a blindfold b...
http://arxiv.org/pdf/1811.05013v1
2018-11-12T21:45:41
2018-11-12T21:45:41
{ "title": [ -0.03286743, -0.0017967224, -0.030380249, -0.016418457, -0.021209717, -0.012886047, -0.0078125, -0.009933472, 0.018585205, 0.037475586, -0.04638672, 0.007835388, 0.021392822, -0.0035552979, -0.019577026, 0.0090789795, 0.0076942444, 0...
http://arxiv.org/abs/2307.00028v1
Seeing in Words: Learning to Classify through Language Bottlenecks
Khalid Saifullah, Yuxin Wen, Jonas Geiping, Micah Goldblum, Tom Goldstein
cs.CV, cs.AI, cs.CL, cs.LG
Neural networks for computer vision extract uninterpretable features despite achieving high accuracy on benchmarks. In contrast, humans can explain their predictions using succinct and intuitive descriptions. To incorporate explainability into neural networks, we train a vision model whose feature representations are t...
http://arxiv.org/pdf/2307.00028v1
2023-06-29T00:24:42
2023-06-29T00:24:42
{ "title": [ -0.024002075, -0.0041503906, -0.030715942, -0.033996582, 0.0096588135, -0.009849548, -0.016784668, -0.012840271, 0.009010315, 0.011856079, -0.043945312, 0.012832642, -0.020629883, -0.004711151, -0.0491333, -0.011367798, -0.005760193, ...
http://arxiv.org/abs/1909.07583v2
Inverse Visual Question Answering with Multi-Level Attentions
Yaser Alwattar, Yuhong Guo
cs.CV, cs.AI, cs.CL, cs.LG
In this paper, we propose a novel deep multi-level attention model to address inverse visual question answering. The proposed model generates regional visual and semantic features at the object level and then enhances them with the answer cue by using attention mechanisms. Two levels of multiple attentions are employed...
http://arxiv.org/pdf/1909.07583v2
2020-12-03T00:13:21
2019-09-17T04:41:12
{ "title": [ 0.00037264824, -0.003929138, -0.010604858, -0.03375244, -0.011070251, -0.005180359, -0.007633209, -0.005760193, 0.045928955, 0.024856567, -0.0340271, -0.010925293, 0.008666992, 0.00016701221, -0.057434082, -0.002117157, 0.010520935, ...
http://arxiv.org/abs/1508.05508v1
Towards Neural Network-based Reasoning
Baolin Peng, Zhengdong Lu, Hang Li, Kam-Fai Wong
cs.AI, cs.CL, cs.LG, cs.NE
We propose Neural Reasoner, a framework for neural network-based reasoning over natural language sentences. Given a question, Neural Reasoner can infer over multiple supporting facts and find an answer to the question in specific forms. Neural Reasoner has 1) a specific interaction-pooling mechanism, allowing it to exa...
http://arxiv.org/pdf/1508.05508v1
2015-08-22T13:15:09
2015-08-22T13:15:09
{ "title": [ -0.030090332, -0.0046424866, -0.032348633, -0.05117798, 0.017471313, 0.009185791, -0.020233154, -0.00012028217, 0.011207581, 0.017807007, -0.017669678, -0.022735596, -0.019104004, -0.028411865, -0.04638672, -0.0019140244, -0.0018978119, ...