16,450 Hits in 7.3 sec

EXIST 2021: Inducing Bias in Deep Learning Models

Ismael Garrido-Muñoz, Arturo Montejo-Ráez, Fernando Javier Martínez Santiago
2021 Annual Conference of the Spanish Society for Natural Language Processing  
For the EXIST 2021 sexism detection task, a novel approach is proposed. In previous work we see that deep learning models capture biases related to race, gender, religion, etc.  ...  There are numerous works that seek to lower the bias of these models to make them fairer and more equitable.  ...  The study of bias in neural network based models shows how recent models such as BERT or even GPT-3 capture these biases.  ... 
dblp:conf/sepln/Garrido-MunozMS21 fatcat:oxfqfzzaargixporatsvj4rsyy

Bad Global Minima Exist and SGD Can Reach Them [article]

Shengchao Liu, Dimitris Papailiopoulos, Dimitris Achlioptas
2021 arXiv   pre-print
We take a careful look at this explanation in the context of image classification with common deep neural network architectures.  ...  The consensus explanation that has emerged credits the randomized nature of SGD for the bias of the training process towards low-complexity models and, thus, for implicit regularization.  ...  In contrast, we find that the bias towards simple models induced by regularization and data augmentation, does.  ... 
arXiv:1906.02613v2 fatcat:bh3pgrt3jvddxhknmreimoipl4

Software Architecture for ML-based Systems: What Exists and What Lies Ahead [article]

Henry Muccini, Karthik Vaidhyanathan
2021 arXiv   pre-print
In this work, we focus on the former side of the spectrum with a goal to highlight the different architecting practices that exist in the current scenario for architecting ML-based software systems.  ...  The increasing usage of machine learning (ML) coupled with the software architectural challenges of the modern era has resulted in two broad research areas: i) software architecture for ML-based systems  ...  A complex deep learning algorithm might produce a heavier and highly accurate model (e.g., a model with more neural network layers).  ... 
arXiv:2103.07950v2 fatcat:dtybnizkavge7cv6wif3qe7p4m

Deep Learning Point Cloud Odometry

Bernardo Teixeira, Hugo Silva
2021 U Porto Journal of Engineering  
comparison.Furthermore, existing Deep Learning approaches and techniques for point cloud odometry tasks are explored, and the main technological solutions are compared and discussed.Open challenges are  ...  In addition, a broad perspective of emergent Deep Learning robust techniques to track motion and estimate scene structure for real-world applications is the focus of a deeper analysis and comprehensive  ...  , delving deeper on deep learning (DL) itself and the huge advances on learning-based models for Computer Vision; Section 3 compares existing approaches to DL-based LIDAR odometry estimation, zooming in  ... 
doi:10.24840/2183-6493_007.003_0006 fatcat:orrlfkowkzgsjijnl5vnlczzzm

On the Existence of Universal Lottery Tickets [article]

Rebekka Burkholz, Nilanjana Laha, Rajarshi Mukherjee, Alkis Gotovos
2022 arXiv   pre-print
The lottery ticket hypothesis conjectures the existence of sparse subnetworks of large randomly initialized deep neural networks that can be successfully trained in isolation.  ...  Our explicit sparse constructions of universal function families might be of independent interest, as they highlight representational benefits induced by univariate convolutional architectures.  ...  applications, a lack of fairness due to biases in data, or vulnerabilities of deployed models to adversarial examples.  ... 
arXiv:2111.11146v2 fatcat:4mfefveicfd2zir6orjoiqa7du

Verifying the existence of maximum likelihood estimates for generalized linear models [article]

Sergio Correia, Paulo Guimarães, Thomas Zylkin
2021 arXiv   pre-print
A fundamental problem with nonlinear models is that maximum likelihood estimates are not guaranteed to exist.  ...  Though nonexistence is a well known problem in the binary choice literature, it presents significant challenges for other models as well and is not as well understood in more general settings.  ...  Z (2021): "Bias and Consistency in Three-Way Gravity Models," unpublished manuscript. W , R. (2008): Econometric Analysis of Count Data, Springer Science & Business Media. Y , Y. V., R. P , J. A.  ... 
arXiv:1903.01633v6 fatcat:bm64e5nigbfetel3whhjvargbq

Structures and Forms of Existence [chapter]

2021 The Pechenegs: Nomads in the Political and Cultural Landscape of Medieval Europe  
We learn about their existence from the accounts of both Anna Komnena and John Skylitzes. 320 for failure to engage in combat, insubordination or desertion?  ...  If so, then Marey's reconstructed model of the Pechenegs' political structure would be identical to the segmentary model.  ... 
doi:10.1163/9789004441095_006 fatcat:5f5b5lf6dzfs7a5ttfxudlh6pe

Advances in MetaDL: AAAI 2021 challenge and workshop [article]

Adrian El Baz, Isabelle Guyon
2022 arXiv   pre-print
To stimulate advances in metalearning using deep learning techniques (MetaDL), we organized in 2021 a challenge and an associated workshop.  ...  The challenge focused on few-shot learning classification tasks of small images. Participants' code submissions were run in a uniform manner, under tight computational constraints.  ...  Meskhi et al. ( 2021 ) propose a novel method of inducing abstract meta-features as latent variables learned by a deep neural network.  ... 
arXiv:2202.01890v1 fatcat:ctadmutvujhizanao5kfuiu2ja

Evo* 2021 – Late-Breaking Abstracts Volume [article]

A.M. Mora, A.I. Esparcia-Alcázar
2021 arXiv   pre-print
Volume with the Late-Breaking Abstracts submitted to the Evo* 2021 Conference, held online from 7 to 9 of April 2021.  ...  Acknowledgements Berker Banar is a PhD student at the UKRI Centre for Doctoral Training in Artificial Intelligence and Music, funded by UKRI grant number EP/S022694/1 and Queen Mary University of London  ...  Keywords: Evolutionary Computation · Neural Networks · Deep Neu-roEvolution. Introduction The Machine Learning community has come a long way since the first deep learning models were proposed.  ... 
arXiv:2106.11804v1 fatcat:6otgwnlqsfev5jt4fnux32ffly

Guidelines for energy efficiency HBIM development of existing buildings

Elena Gigliarelli, Filippo Calcerano, Letizia Martinelli, Stavroula Thravalou, Kristis Alexandrou, Georgios Artopoulos
2021 Zenodo  
and performance-based design to develop an Energy Efficient Heritage Building Information Model - EE-HBIM).  ...  This pdf provides a technical guideline for an energy audit of a historical building to support its energy and environmental improvement (as shown in the Energy Audit Process Flow schema based on the EN  ...  In Italy from 1 January 2018, new buildings and those subjected to a deep renovation (for existing buildings with surfaces exceeding 100sqm and subject to complete renovation of the building elements making  ... 
doi:10.5281/zenodo.6393028 fatcat:7kpyhh336jaxfgt6ocu4rdaca4

ISDCS 2021 Technical Program

2021 2021 International Symposium on Devices, Circuits and Systems (ISDCS)  
ISDCS 2021 will be held online with home base in Hiroshima University, Higashi-Hiroshima, Japan from March 3 to 5.  ...  This year, the ISDCS Program Committee received a total of fine submissions from 7 countries, of which 32 outstanding contributions would be included in the program of ISDCS 2021.  ...  Due to the huge bias applied, a new leakage-current type is induced during switching, which causes a switching-power-loss increase.  ... 
doi:10.1109/isdcs52006.2021.9397912 fatcat:766wjfco3fe2fgat2i42etyk5q

Understanding the complexity of existing fossil fuel power plant decarbonization

Chuan Zhang, Haibo Zhai, Liwei Cao, Xiang Li, Fangwei Cheng, Liqun Peng, Kangkang Tong, Jing Meng, Lei Yang, Xiaonan Wang
2022 iScience  
Growing national decarbonization commitments require rapid and deep reductions of carbon dioxide emissions from existing fossil-fuel power plants.  ...  fossil-fuel power plant decarbonization challenges that imposes the need for a serious re-evaluation of existing fossil fuel power plant retrofit in energy transition.  ...  models (Meng et al., 2021) .  ... 
doi:10.1016/j.isci.2022.104758 fatcat:aynkigwo75bfldbr53afprirxm

Eastern Regional Meeting 2021

2021 Journal of Investigative Medicine  
SB203580 failed to prevent DOX-induced CV slowing in mice of both sexes (19% and 18% CV reduction in males and females, respectively).  ...  We performed landmark Kaplan Meier analysis for survival and recurrence, and times were offset by six months to correct for immortal time bias.  ...  The machine learning model was engineered using deep learning algorithms fitted with neural networks.  ... 
doi:10.1136/jim-2021-erm fatcat:xdufm735hvadxnmnqeutlhigce

The AI Index 2021 Annual Report [article]

Daniel Zhang, Saurabh Mishra, Erik Brynjolfsson, John Etchemendy, Deep Ganguli, Barbara Grosz, Terah Lyons, James Manyika, Juan Carlos Niebles, Michael Sellitto, Yoav Shoham, Jack Clark (+1 others)
2021 arXiv   pre-print
The report aims to be the most credible and authoritative source for data and insights about AI in the world.  ...  This year we significantly expanded the amount of data available in the report, worked with a broader set of external organizations to calibrate our data, and deepened our connections with the Stanford  ...  The terms searched for were based on the issues exposed and identified in papers below, and also on the topics called for discussion in the First AAAI/ACM Conference on AI, Ethics, and Society.  ... 
arXiv:2103.06312v1 fatcat:52qwvzv7jndxzaagyiro6koyza

Proceedings of the 2021 ISNR Annual Conference (Virtual): Poster Presentations

International Society of NeuroRegulation and Research (ISNR)
2021 NeuroRegulation  
The deep brain activity at the frontal lobe and PCC in different frequency bands were analyzed for delta, theta, alpha, and beta. Results.  ...  This study aimed to convert EEG raw signals into swLORETA and explore the differences in deep brain activity between patients with MDD and healthy adults. Method.  ...  Craving and attentional bias towards drug-related items may contribute to the high rates of noncompliance and relapse in OUD individuals undergoing MAT.  ... 
doi:10.15540/nr.8.4.220 fatcat:ntdmqappcbejzc5i3nrfxvutua
« Previous Showing results 1 — 15 out of 16,450 results