Filters








26 Hits in 3.7 sec

Queried Unlabeled Data Improves and Robustifies Class-Incremental Learning [article]

Tianlong Chen, Sijia Liu, Shiyu Chang, Lisa Amini, Zhangyang Wang
2022 arXiv   pre-print
Class-incremental learning (CIL) suffers from the notorious dilemma between learning newly added classes and preserving previously learned class knowledge.  ...  Along with new and past stored data, the queried unlabeled are effectively utilized, through learning-without-forgetting (LwF) regularizers and class-balance training.  ...  Army Research Laboratory Cooperative Research Agreement W911NF17-2-0196 (IOBT REIGN), and an IBM faculty research award.  ... 
arXiv:2206.07842v2 fatcat:fxjmwg744nb2pbd3x6bn4cupbe

Adaptive-Attentive Geolocalization From Few Queries: A Hybrid Approach

Valerio Paolicelli, Gabriele Berton, Francesco Montagna, Carlo Masone, Barbara Caputo
2022 Frontiers in Computer Science  
In particular, at training time, we consider having access to only few unlabeled queries from the target domain.  ...  We tackle the task of cross-domain visual geo-localization, where the goal is to geo-localize a given query image against a database of geo-tagged images, in the case where the query and the database belong  ...  data augmentation, attention, domain adaptation) provide orthogonal increments to the performance.  ... 
doi:10.3389/fcomp.2022.841817 fatcat:4b2rjdzhxbenzo3nte2k2iiv5u

Noisy Batch Active Learning with Deterministic Annealing [article]

Gaurav Gupta, Anit Kumar Sahu, Wan-Yi Lin
2020 arXiv   pre-print
We introduce an extra denoising layer to deep networks to make active learning robust to label noises and show significant improvements.  ...  Experiments on benchmark image classification datasets (MNIST, SVHN, CIFAR10, and EMNIST) show improvement over existing active learning strategies.  ...  data is too small to build a meaningful model; (3) we introduce a denoising layer to deep networks to robustify active learning to noisy oracles.  ... 
arXiv:1909.12473v2 fatcat:6msrfnuwi5hb3a2hoco3s2w27u

Adaptive Risk Minimization: Learning to Adapt to Domain Shift [article]

Marvin Zhang, Henrik Marklund, Nikita Dhawan, Abhishek Gupta, Sergey Levine, Chelsea Finn
2021 arXiv   pre-print
A fundamental assumption of most machine learning algorithms is that the training and test data are drawn from the same underlying distribution.  ...  In contrast, we aim to learn models that adapt at test time to domain shift using unlabeled test points.  ...  Acknowledgments and Disclosure of Funding  ... 
arXiv:2007.02931v4 fatcat:2gjsygbev5cmhdryhrwijvfeju

Content-based visual search learned from social media

Xirong Li
2012 ACM SIGMultimedia Records  
, Gosia and Hamdi for the ISIS dinners, Daan and Bouke for sharing conference hotels, and our secretary Virginie who helped me much in the past years.  ...  It is my pleasure to work in the ISIS group: Michael, Ivo, and Jan as my ping pong mates for years, Victoria, Vladimir, Dung, and Stratios for sharing offices, Cor, Jan-Mark, and Theo for interesting discussions  ...  For most of the 20 queries, we improve upon the baseline methods by using learned tag frequency as updated tag frequency.  ... 
doi:10.1145/2206765.2206774 fatcat:kxk6kciwhfe2hcqw546ez2coku

Meta-Learning in Neural Networks: A Survey [article]

Timothy Hospedales, Antreas Antoniou, Paul Micaelli, Amos Storkey
2020 arXiv   pre-print
This paradigm provides an opportunity to tackle many conventional challenges of deep learning, including data and computation bottlenecks, as well as generalization.  ...  Contrary to conventional approaches to AI where tasks are solved from scratch using a fixed learning algorithm, meta-learning aims to improve the learning algorithm itself, given the experience of multiple  ...  Current research issues include improving cross-domain generalization [119] , recognition within the joint label space defined by metatrain and meta-test classes [84] , and incremental addition of new  ... 
arXiv:2004.05439v2 fatcat:3r23tsxxkfbgzamow5miglkrye

SETTI: A S elf-Supervised Adv E rsarial Malware De T ection Archi T ecture in an I oT Environment

Marjan Golmaryami, Rahim Taheri, Zahra Pooranian, Mohammad Shojafar, Pei Xiao
2022 ACM Transactions on Multimedia Computing, Communications, and Applications (TOMCCAP)  
The Self-MDS method considers the IoT input data and the adversarial sample generation in real-time.  ...  More research is thus required on malware detection to cope with real-time misclassification of the input IoT data.  ...  DETECTIONSelf-supervised Learning by rote is an unsupervised learning technique that involves creating a supervised learning task from unlabeled input data.  ... 
doi:10.1145/3536425 fatcat:noskwsordnehrjqcjabsqmk47m

SETTI: A Self-supervised Adversarial Malware Detection Architecture in an IoT Environment [article]

Marjan Golmaryami, Rahim Taheri, Zahra Pooranian, Mohammad Shojafar, Pei Xiao
2022 arXiv   pre-print
The Self-MDS method considers the IoT input data and the adversarial sample generation in real-time.  ...  More research is thus required on malware detection to cope with real-time misclassification of the input IoT data.  ...  SELF-SUPERVISED MALWARE DETECTION Self-supervised Learning by rote is an unsupervised learning technique that involves creating a supervised learning task from unlabeled input data.  ... 
arXiv:2204.07772v1 fatcat:v6pmiwkbnngd3kkmbhsrj46csq

Meta-Learning in Neural Networks: A Survey

Timothy M Hospedales, Antreas Antoniou, Paul Micaelli, Amos J. Storkey
2021 IEEE Transactions on Pattern Analysis and Machine Intelligence  
This paradigm provides an opportunity to tackle many of the conventional challenges of deep learning, including data and computation bottlenecks, as well as the fundamental issue of generalization.  ...  Contrary to conventional approaches to AI where a given task is solved from scratch using a fixed learning algorithm, meta-learning aims to improve the learning algorithm itself, given the experience of  ...  Current research issues include improving cross-domain generalization [115] , recognition within the joint label space defined by metatrain and meta-test classes [81] , and incremental addition of new  ... 
doi:10.1109/tpami.2021.3079209 pmid:33974543 fatcat:wkzeodki4fbcnjlcczn4mr6kry

Advances in adversarial attacks and defenses in computer vision: A survey [article]

Naveed Akhtar, Ajmal Mian, Navid Kardan, Mubarak Shah
2021 arXiv   pre-print
To ensure authenticity, we mainly consider peer-reviewed contributions published in the prestigious sources of computer vision and machine learning research.  ...  In [2], we reviewed the contributions made by the computer vision community in adversarial attacks on deep learning (and their defenses) until the advent of year 2018.  ...  Similarly, to improve query efficiency, a technique to extract generalizable prior using the earlier queries with meta learning is proposed in [78] .  ... 
arXiv:2108.00401v2 fatcat:23gw74oj6bblnpbpeacpg3hq5y

The Neural Process Family: Survey, Applications and Perspectives [article]

Saurav Jha, Dong Gong, Xuesong Wang, Richard E. Turner, Lina Yao
2022 arXiv   pre-print
We then provide a rigorous taxonomy of the family and empirically demonstrate their capabilities for modeling data generating functions operating on 1-d, 2-d, and 3-d input domains.  ...  Gaussian processes, on the other hand, adopt the Bayesian learning scheme to estimate such uncertainties but are constrained by their efficiency and approximation capacity.  ...  of total classes to form C and T accounts for the improved generalizability of the learner.  ... 
arXiv:2209.00517v1 fatcat:2vxwayau35ctdehgl22yajxvri

Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety [article]

Sebastian Houben, Stephanie Abrecht, Maram Akila, Andreas Bär, Felix Brockherde, Patrick Feifel, Tim Fingscheidt, Sujan Sai Gannamaneni, Seyed Eghbal Ghobadi, Ahmed Hammam, Anselm Haselhoff, Felix Hauser (+29 others)
2021 arXiv   pre-print
Our paper addresses both machine learning experts and safety engineers: The former ones might profit from the broad range of machine learning topics covered and discussions on limitations of recent methods  ...  This work provides a structured and broad overview of them.  ...  Furthermore, this research has been funded by the Federal Ministry of Education and Research of Germany as part of the competence center for machine learning ML2R (01IS18038B).  ... 
arXiv:2104.14235v1 fatcat:f6sj3v2brza7thyzw7b7fkpo2m

Convergence of Photovoltaic Power Forecasting and Deep Learning: State-of-Art Review

Mohamed Massaoudi, Ines Chihi, Haitham Abu-Rub, Shady S. Refaat, Fakhreddine S. Oueslati
2021 IEEE Access  
INCREMENTAL AND ONLINE LEARNING Under the umbrella of DL, incremental and online Leaning has emerged as a continuous evolving scheme to improve the universality of the prediction engines for accurate regional  ...  ) architecture is one of the most groundbreaking unsupervised learning models that learn characteristics from unlabeled data representation [9] .  ... 
doi:10.1109/access.2021.3117004 fatcat:nxgyb5e4rvbynpmgzppv7ecbre

Pain Intensity Estimation by a Self--Taught Selection of Histograms of Topographical Features [article]

Corneliu Florea, Laura Florea, Raluca Boia, Alessandra Bandrabur, Constantin Vertan
2015 arXiv   pre-print
We use this procedure to improve the discrimination between different pain intensity levels and the generalization with respect to the monitored persons, while testing on the UNBC McMaster Shoulder Pain  ...  We propose a semi-supervised, clustering oriented self--taught learning procedure developed on the emotion oriented Cohn-Kanade database.  ...  The work has been partially funded by the Sectoral Operational Programme Human Resources Development 2007-2013 of the Ministry of European Funds through the Financial Agreement POSDRU/159/1.5/S/ 134398 and  ... 
arXiv:1503.07706v1 fatcat:witmehll7vfinfuo47h3uancmi

Deep Neural Mobile Networking [article]

Chaoyun Zhang
2020 arXiv   pre-print
In particular, deep learning based solutions can automatically extract features from raw data, without human expertise.  ...  This makes monitoring and managing the multitude of network elements intractable with existing tools and impractical for traditional machine learning algorithms that rely on hand-crafted feature engineering  ...  Classifier robustifying employs various approaches (e.g., model ensembling) to improve the robustness of the original classifier.  ... 
arXiv:2011.05267v1 fatcat:yz2zp5hplzfy7h5kptmho7mbhe
« Previous Showing results 1 — 15 out of 26 results