Filters








502 Hits in 3.6 sec

A Relation Learning Hierarchical Framework for Multi-label Charge Prediction [chapter]

Wei Duan, Lin Li, Yi Yu
2020 Lecture Notes in Computer Science  
NLN mitigates the dynamic label number by learning the co-occurring relation between labels. Moreover, we put the two models into a unified framework to enhance their effects.  ...  In legal field, multi-label charge prediction is a popular and foundational task to predict charges (labels) by a case description (a fact).  ...  For a fact X, using a feature extractor can gain its knowledge-free representation h 1 . For X and all provisionsX, we get a knowledge-aware representation h 2 by DMA.  ... 
doi:10.1007/978-3-030-47436-2_55 fatcat:oaw4vjt2kjahhpbbghes2myvxi

Slices of Attention in Asynchronous Video Job Interviews [article]

Léo Hemamou, Ghazi Felhi, Jean-Claude Martin, Chloé Clavel
2019 arXiv   pre-print
Second, we study the content of attention slices by comparing them with randomly sampled slices.  ...  However, these studies are still mainly based on hand engineered features, which imposes a limit to the discovery of influential social signals.  ...  ACKNOWLEDGMENT This work was supported by the company EASYRECRUE. We would like to thank Jeremy Langlais and Amandine Reitz for their support and their help.  ... 
arXiv:1909.08845v1 fatcat:vw7coxdyy5hltoylbzwgrm3eka

Slices of Attention in Asynchronous Video Job Interviews

Leo Hemamou, Ghazi Felhi, Jean-Claude Martin, Chloe Clavel
2019 2019 8th International Conference on Affective Computing and Intelligent Interaction (ACII)  
Second, we study the content of attention slices by comparing them with randomly sampled slices.  ...  However, these studies are still mainly based on hand engineered features, which imposes a limit to the discovery of influential social signals.  ...  ACKNOWLEDGMENT This work was supported by the company EASYRECRUE. We would like to thank Jeremy Langlais and Amandine Reitz for their support and their help.  ... 
doi:10.1109/acii.2019.8925439 dblp:conf/acii/HemamouFMC19 fatcat:vlar6u2uyncrbajpmbqlpr4fum

Computational Complexity and the Function-Structure-Environment Loop of the Brain [chapter]

B. Juba
2016 Closed Loop Neuroscience  
Computational complexity is a potentially useful conceptual framework because it enables the meaningful study of the family of possible structures as a whole-the study of "the brain," as opposed to some  ...  The language of computational complexity also provides a means of formally capturing capabilities of the brain, which may otherwise be philosophically thorny.  ...  Acknowledgements I thank Sashank Varma and Frank Jäkel for a series of conversations that shaped the section on identifiability and invariance, and their comments on a previous draft.  ... 
doi:10.1016/b978-0-12-802452-2.00010-x fatcat:rkkz7mzmrbh73e2xatpilouk6u

The Effect of Dwell Time on Swipe-based Pie-Menu Navigation Efficiency

Alen Salkanovic, Matija Stojkovic, Sandi Ljubic
2020 International Journal of Online and Biomedical Engineering (iJOE)  
We propose a solution that synergistically encompasses important design features for a given context: radial menu visualization, semi-transparency, manual repositioning, occlusion awareness, and marking  ...  From the menu navigation efficiency standpoint, dwell time represents one of the most influential factors in such a design.  ...  A pointer controlled by the finger is used to select a certain menu item in order to avoid the occlusion problem.  ... 
doi:10.3991/ijoe.v16i15.17763 fatcat:dtkphe2uuzguzikeczxkvoxbp4

CATs++: Boosting Cost Aggregation with Convolutions and Transformers [article]

Seokju Cho, Sunghwan Hong, Seungryong Kim
2022 arXiv   pre-print
Also, to alleviate some of the limitations that CATs may face, i.e., high computational costs induced by the use of a standard transformer that its complexity grows with the size of spatial and feature  ...  Our proposed methods outperform the previous state-of-the-art methods by large margins, setting a new state-of-the-art for all the benchmarks, including PF-WILLOW, PF-PASCAL, and SPair-71k.  ...  of the positional information by providing a learnable embedding [42] rather than a fixed [41] as shown in Fig. 4 .  ... 
arXiv:2202.06817v1 fatcat:pdwkvhrcwzfzbnjr2nj5w7xkuq

DARPA's Role in Machine Learning

Joshua Alspector, Thomas Dietterich
2020 The AI Magazine  
models, the integration of multiple machine learning approaches into a task-specific system, and neural network technology.  ...  This article describes four threads of machine learning research supported and guided by the Defense Advanced Research Projects Agency — probabilistic modeling for speech recognition, probabilistic relational  ...  Taken from Andrew Ng's talk, unsupervised Feature Learning and Deep Learning www.csee.umbc.edu/courses/ pub/www/courses/graduate/678/fall14/visionaudio.pdf  ... 
doi:10.1609/aimag.v41i2.5298 fatcat:p2tsjhuk6fd6xhp4ttyufsrpay

Influence Maximization on Social Graphs: A Survey

Yuchen Li, Ju Fan, Yanhao Wang, Kian-Lee Tan
2018 IEEE Transactions on Knowledge and Data Engineering  
Influence Maximization (IM), which selects a set of k users (called seed set) from a social network to maximize the expected number of influenced users (called influence spread), is a key algorithmic problem  ...  models that capture information diffusion process and build the foundation of the IM problem, (2) a fine-grained taxonomy to classify existing IM algorithms based on their design objectives, (3) a rigorous  ...  The Linear Threshold (LT) Model Linear Threshold (LT) is also a seminal diffusion model, which is introduced by Granovetter and Schelling [39] , [94] in 1978.  ... 
doi:10.1109/tkde.2018.2807843 fatcat:f6vnhknlevckbkqwkkqotgalc4

The Utility of Abstaining in Binary Classification [article]

Akshay Balsubramani
2015 arXiv   pre-print
This is directly motivated by applications like medical diagnosis and fraud risk assessment, in which incorrect predictions have potentially calamitous consequences.  ...  We focus on a recent spate of theoretically driven work in this area that characterizes how allowing abstentions can lead to fewer errors in very general settings.  ...  Abstaining in the KWIK Model In this section, we describe a general model which has been influential in recent advances in the abstention literature.  ... 
arXiv:1512.08133v1 fatcat:q7ekqcjdsnahnjmgpgpzdwqrba

Intelligence, physics and information – the tradeoff between accuracy and simplicity in machine learning [article]

Tailin Wu
2020 arXiv   pre-print
Inspired by how physicists model the world, we introduce a paradigm and an AI Physicist agent for simultaneously learning many small specialized models (theories) and the domain they are accurate, which  ...  Fourthly, to make models more robust to label noise, we introduce Rank Pruning, a robust algorithm for classification with noisy labels.  ...  Another closely related field is sparse learning/feature selection methods.  ... 
arXiv:2001.03780v2 fatcat:piduzlhoafcjhhsgthulbbhtke

The impact of video encoding parameters and game type on QoE for cloud gaming: A case study using the steam platform

Ivan Slivar, Mirko Suznjevic, Lea Skorin-Kapov
2015 2015 Seventh International Workshop on Quality of Multimedia Experience (QoMEX)  
We test our hypothesis through a series of subjective tests on the streaming solution of the Steam In-home streaming platform.  ...  results has received funding from the project "Information and communication technology for generic and energy-efficient communication solutions with application in e-/m-health (ICTGEN))" co-financed by  ...  configurations QoE features for both games, as shown in Fig. 4 . 4 Summary of linear regression models for QoE with video parameters and QoE features Fig. 5.  ... 
doi:10.1109/qomex.2015.7148144 dblp:conf/qomex/SlivarSS15 fatcat:slt7dxrbjnc4re7pzumyopgf2y

A Spatial-temporal Graph Deep Learning Model for Urban Flood Nowcasting Leveraging Heterogeneous Community Features [article]

Hamed Farahmand, Yuanchang Xu, Ali Mostafavi
2021 arXiv   pre-print
The objective of this study is to develop and test a novel structured deep-learning modeling framework for urban flood nowcasting by integrating physics-based and human-sensed features.  ...  Third, its attention mechanism enables the model to direct its focus on the most influential features that vary dynamically.  ...  Systems Resilience to Urban Flooding: Integrated Assessment of Social, Institutional, and Physical Networks and the funding support from the X-Grant program (Presidential Excellence Fund) from Texas A&  ... 
arXiv:2111.08450v2 fatcat:6ockrwwz7fgfnhnmaynmcgtwwy

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, Alan L. Yuille
2018 IEEE Transactions on Pattern Analysis and Machine Intelligence  
We incorporate such masks in CNNs and replace the convolution operation with a "segmentationaware" variant that allows a neuron to selectively attend to inputs coming from its own region.  ...  We call the resulting network a segmentation-aware CNN because it adapts its filters at each image point according to local segmentation cues.  ...  behavior to the image content.  ... 
doi:10.1109/tpami.2017.2699184 pmid:28463186 fatcat:44cino7l4jfjngddvqw44srupm

Deep Learning on Knowledge Graph for Recommender System: A Survey [article]

Yang Gao, Yi-Fan Li, Yu Lin, Hang Gao, Latifur Khan
2020 arXiv   pre-print
In this paper, we provide a comprehensive survey of the GNN-based knowledge-aware deep recommender systems.  ...  A knowledge graph is capable of encoding high-order relations that connect two objects with one or multiple related attributes.  ...  By sampling a fixed number of neighboring nodes, a node may lose its influential neighbors.  ... 
arXiv:2004.00387v1 fatcat:vowfwf7dhnau7nzrlnkjl3a2de

DeepCP: Deep Learning Driven Cascade Prediction Based Autonomous Content Placement in Closed Social Network [article]

Qiong Wu and Muhong Wu and Xu Chen and Zhi Zhou and Kaiwen He and Liang Chen
2020 arXiv   pre-print
We first devise a time-window LSTM model for content popularity prediction and cascade geo-distribution estimation.  ...  In this paper, we take a new direction of popularity-aware content placement in a closed social network (e.g., WeChat Moment) where user's privacy is highly enhanced.  ...  [11] represent a cascade by five classes of features (content features, original poster features, resharer features, structural features and temporal features) and then use machine learning classifiers  ... 
arXiv:2003.03971v1 fatcat:xnldollrkfblna3bk5n4lqgwgy
« Previous Showing results 1 — 15 out of 502 results