8 Hits in 6.1 sec

Hangul Fonts Dataset: a Hierarchical and Compositional Dataset for Investigating Learned Representations [article]

Jesse A. Livezey, Ahyeon Hwang, Jacob Yeung, Kristofer E. Bouchard
2021 arXiv   pre-print
To address this gap, we developed a new benchmark dataset with known hierarchical and compositional structure.  ...  Determining when a deep network's hidden activations represent hierarchy and compositionality is important both for understanding deep representation learning and for applying deep networks in domains  ...  Acknowledgments and Disclosure of Funding JAL, AH, and KEB were supported by the Deep Learning for Science Laboratory Directed Research and Development.  ... 
arXiv:1905.13308v2 fatcat:tfoqcr3p25hsriqx7u2oa4pzna

Synbols: Probing Learning Algorithms with Synthetic Datasets [article]

Alexandre Lacoste, Pau Rodríguez, Frédéric Branchaud-Charron, Parmida Atighehchian, Massimo Caccia, Issam Laradji, Alexandre Drouin, Matt Craddock, Laurent Charlin, David Vázquez
2020 arXiv   pre-print
In this sense, we introduce Synbols -- Synthetic Symbols -- a tool for rapidly generating new datasets with a rich composition of latent features rendered in low resolution images.  ...  distribution generalization, unsupervised representation learning, and object counting.  ...  Acknowledgments and Disclosure of Funding This work was majorly supported by Element AI. Laurent Charlin is supported through a CIFAR AI Chair and grants from NSERC, CIFAR, IVADO, Samsung, and Google.  ... 
arXiv:2009.06415v2 fatcat:tbqqsdrfwfcv5nnhl3h4re6cxa

Handwritten Bangla Digit Recognition Using Deep Learning [article]

Md Zahangir Alom, Paheding Sidike, Tarek M. Taha, Vijayan K. Asari
2017 arXiv   pre-print
recognition and machine learning applications, but has not been throughly attempted for HBDR.  ...  These networks have the advantage of extracting and using feature information, improving the recognition of two dimensional shapes with a high degree of invariance to translation, scaling and other pattern  ...  Experimental results and discussion Dataset description We evaluated the performance of DBN and CNN on a benchmark dataset called CMATERdb 3.1.1 (Das et al., 2012a; b) .  ... 
arXiv:1705.02680v1 fatcat:w6vjh4hkv5dbdhsi7qdcrcls7a

Final Program, Thirty-Sixth Annual Meeting International Neuropsychological Society, February 6-9, 2008, Waikoloa, Hawaii, USA

2008 Journal of the International Neuropsychological Society  
We also investigated the effects of side of PD onset upon the spatial representation of number.  ...  A hierarchical regression found that a phonological factor accounted for the most variance in the model (R2 = .221; p = .009).  ...  Such methods may be essential for analyses of large multicenter datasets and may be relevant for future clinical classifications.  ... 
doi:10.1017/s1355617708080429 fatcat:7cinmclodjdjdfwakkdssx2iiy

Effects of Implicit and Explicit Focus on Form on L2 Acquisition of the English Passive

Ji-Yung Jung
Second, the difference appeared to be more salient for meaning and function, whereas both types of instruction had [...]  ...  To address this gap, the present study investigated the effects of implicit and explicit instruction on L2 acquisition of the English passive.  ...  The dimension of L1 markedness is conceived of as a composite of two sub-variables, frequency and variability (Panel a in Figure 4 ).  ... 
doi:10.7916/d8-ek22-ka98 fatcat:p467fed7hjgldbynfcd65uouyq

Exploring Naming Behavior in Personal Digital Image Collections

Tami Sutcliffe
2016 Leonardo: Journal of the International Society for the Arts, Sciences and Technology  
In order to capture and explore this evolving language adaptation, Pinterest names are analyzed using a matrix composed of Panofsky's three strata of subject matter, Rosch's levels of categorical abstraction  ...  , Shatford Layne's image attributes and Wittgenstein's language game constructions.  ...  lower case fonts abbreviations, and malformed sentence/word phrases 6.  ... 
doi:10.1162/leon_a_01304 fatcat:ulj4k2rjbbbwtfts6jyczkpvsq

On the Origin of Arabic Script

Kamal Mansour
Graphemics in the 21st Century   unpublished
Not only have I greatly profited from our discussions about Chinese scripts, the character system, cal ligraphy and so forth, she also always finds a way to get a book or other material for me and to me  ...  , to Mike Cahill, Dave Rowe and Hugh Paterson III for their helpful comments on a draft of this paper, and to the Focolari community in Man for their generous hospitality.  ...  As a result, readers and writers of Chinese must learn three shapes instead of just one for "heart".  ... 
doi:10.36824/2018-graf-mans fatcat:mlqnilf5y5aylimzundwaghswe

Central Asian and Yanbian Korean in comparative perspective

Simon George Barnes-Sadler
affected by the data gathering procedures and representation in Hangul transcription.  ...  is consistent in terms of the amount of data and its precise composition.  ... 
doi:10.25501/soas.00026672 fatcat:fqtpwyzj7nftfhuxvvn6cxviyi