Filters








2,975 Hits in 5.4 sec

Combating False Negatives in Adversarial Imitation Learning [article]

Konrad Zolna, Chitwan Saharia, Leonard Boussioux, David Yu-Tung Hui, Maxime Chevalier-Boisvert, Dzmitry Bahdanau, Yoshua Bengio
2020 arXiv   pre-print
In adversarial imitation learning, a discriminator is trained to differentiate agent episodes from expert demonstrations representing the desired behavior.  ...  We show experimental evidence for this hypothesis and that the 'False Negatives' (i.e. successful agent episodes) significantly hinder adversarial imitation learning, which is the first contribution of  ...  In this work we consider Behavioral Cloning (BC, (Pomerleau 1989) ) as a baseline and our method improves Generative Adversarial Imitation Learning (GAIL) ( False Negatives We found that the False Negative  ... 
arXiv:2002.00412v1 fatcat:aoprz6gpyfhmzo3axf5xuntv4m

Combating False Negatives in Adversarial Imitation Learning (Student Abstract)

Konrad Żołna, Chitwan Saharia, Leonard Boussioux, David Yu-Tung Hui, Maxime Chevalier-Boisvert, Dzmitry Bahdanau, Yoshua Bengio
2020 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
We define the False Negatives problem and show that it is a significant limitation in adversarial imitation learning.  ...  The method, dubbed Fake Conditioning, is tested on instruction following tasks in BabyAI environments, where it improves sample efficiency over the baselines by at least an order of magnitude.  ...  Another IL method, Generative Adversarial Imitation Learning (GAIL) (Ho and Ermon 2016) , has yielded some success by jointly learning reward functions and training policies.  ... 
doi:10.1609/aaai.v34i10.7272 fatcat:uwtbfgmgjrcu3ntbhw65eewmna

Information Stealing in Federated Learning Systems Based on Generative Adversarial Networks [article]

Yuwei Sun, Ng Chong, Hideya Ochiai
2021 arXiv   pre-print
Moreover, by measuring the Euclidean distance between the real data and the reconstructed adversarial samples, we evaluated the performance of the adversary in the learning processes in various scenarios  ...  In this research, we successfully mounted adversarial attacks on a federated learning (FL) environment using three different datasets.  ...  TN (True Negatives) and FN (False Negatives) indicate the number of irrelevant images correctly and incorrectly classified, respectively.(a) Systems initializing.  ... 
arXiv:2108.00701v1 fatcat:emqlmmwilzbcllupzm2z3maw3m

Use of Interpersonal Deception Theory in Counter Social Engineering

Grace Hui Yang, Yue Yu
2018 International Conference on Information and Knowledge Management  
In order to mitigate and combat these attacks, we need better automated counter social engineering algorithms and tools.  ...  In this position paper, we propose a reinforcement learning framework that incorporates interpersonal deception theory to fight against social engineering attacks on social media sites.  ...  Any opinions, findings, conclusions, or recommendations expressed in this paper are of the authors, and do not necessarily reflect those of the sponsor.  ... 
dblp:conf/cikm/YangY18 fatcat:foj7z6tvn5dnnghwydumb6cbcq

Robust Antijamming Strategy Design for Frequency-Agile Radar against Main Lobe Jamming

Kang Li, Bo Jiu, Hongwei Liu, Wenqiang Pu
2021 Remote Sensing  
Then, the method of imitation learning-based jamming strategy parameterization is presented to express the given jamming strategy mathematically.  ...  To combat main lobe jamming, preventive measures can be applied to radar in advance based on the concept of active antagonism, and efficient antijamming strategies can be designed through reinforcement  ...  Imitation learning can be accomplished through three main approaches, which are the behavior cloning [19] , inverse reinforcement learning (IRL) [24] , and generative adversarial imitation learning (  ... 
doi:10.3390/rs13153043 fatcat:3wghsvqeofey3nobfzwspakgqa

Machine Learning for Security and the Internet of Things: the Good, the Bad, and the Ugly

Fan Liang, William G. Hatcher, Weixian Liao, Weichao Gao, Wei Yu
2019 IEEE Access  
In this paper, we consider the good, the bad, and the ugly use of machine learning for cybersecurity and CPS/IoT.  ...  Simultaneously, dramatic improvements in machine learning and deep neural network architectures have enabled unprecedented analytical capabilities, which we see in increasingly common applications and  ...  The implemented framework was demonstrated to be effective without increasing the false alarm rate. 2) ADVERSARIAL LEARNING IN THE TRAINING AND TESTING PHASE Adversarial learning in the training and  ... 
doi:10.1109/access.2019.2948912 fatcat:wxd6imn62fgufdmfh3gtaijeru

Page 119 of None Vol. , Issue 8 [page]

1817 None  
Every where an impartial and judicious criticism dispels the darkness of those remote ages, and combats false or groundless opinions.  ...  adversary, according to the custom of these animals. The eagle was so impetuous in the attack, that no efforts could oppose him, and he tore the picture in pieces.” SCULPTURE, THE VENETIAN HORSES.  ... 

Using Monte Carlo Tree Search as a Demonstrator within Asynchronous Deep RL [article]

Bilal Kartal, Pablo Hernandez-Leal, Matthew E. Taylor
2018 arXiv   pre-print
Deep reinforcement learning (DRL) has achieved great successes in recent years with the help of novel methods and higher compute power.  ...  In this paper, firstly, we augment Asynchronous Advantage Actor-Critic (A3C) method with a novel self-supervised auxiliary task, i.e.  ...  We consider false positive episodes when our agent gets a reward of +1 because the opponent commits suicide (not due to our agent's combat skill), and false negatives episodes when our agent gets a reward  ... 
arXiv:1812.00045v1 fatcat:u7vv6pzp6rdqzcfqexutlpeqiu

Chinese Typeface Transformation with Hierarchical Adversarial Network [article]

Jie Chang, Yujun Gu, Ya Zhang
2017 arXiv   pre-print
Inspired by the recent advancement in Generative Adversarial Networks (GANs), we propose a Hierarchical Adversarial Network (HAN) for typeface transformation.  ...  In this paper, we explore automated typeface generation through image style transfer which has shown great promise in natural image generation.  ...  groundtruth, we propose a hierarchical discriminator for adversarial learning.  ... 
arXiv:1711.06448v1 fatcat:hh4r5ic6zzdetfqciypfmhseui

Epistemic Defenses against Scientific and Empirical Adversarial AI Attacks

Nadisha-Marie Aliman, Leon Kester
2021 International Joint Conference on Artificial Intelligence  
In this paper, we introduce "scientific and empirical adversarial AI attacks" (SEA AI attacks) as umbrella term for not yet prevalent but technically feasible deliberate malicious acts of specifically  ...  In view of possible socio-psychotechnological impacts, it seems responsible to ponder countermeasures from the onset on and not in hindsight.  ...  consciously create and understand novel yet unknown explanatory knowledge [Deutsch, 2011] -which humans are capable of performing if willing to -cannot be learned by AI systems by mere imitation. methods  ... 
dblp:conf/ijcai/AlimanK21 fatcat:rqnhmsax5neodc5r6le7zspmue

Detection under Privileged Information [article]

Z. Berkay Celik, Patrick McDaniel, Rauf Izmailov, Nicolas Papernot, Ryan Sheatsley, Raquel Alvarez, Ananthram Swami
2018 arXiv   pre-print
In this paper, we consider an alternate learning approach that trains models using "privileged" information--features available at training time but not at runtime--to improve the accuracy and resilience  ...  In particular, we adapt and extend recent advances in knowledge transfer, model influence, and distillation to enable the use of forensic or other data unavailable at runtime in a range of security domains  ...  This positive effect substantially improves both false negative and false positive rates.  ... 
arXiv:1603.09638v4 fatcat:xdwbi7zxg5bifiu2em3wcecu5q

Deep Insights of Deepfake Technology : A Review

Bahar Uddin Mahmud, Afsana Sharmin
2020 Zenodo  
In this paper, a wide range of articles have been examined to understand Deepfake technology more extensively.  ...  As Deepfake content creation involve a high level expertise with combination of several algorithms of deep learning, it seems almost real and genuine and difficult to differentiate.  ...  Deepfake Deepfake, a mixtures of deep learning and fake, are imitating contents where targeted subject"s face was swapped by source person to make videos or images of target person [20] [21] .  ... 
doi:10.5281/zenodo.4731421 fatcat:l75l43uwdvbndlb3kbg6qtjgwe

Fused Deep Convolutional Neural Network for Precision Diagnosis of COVID-19 Using Chest X-Ray Images [article]

Hussin K. Ragb, Ian T. Dover, Redha Ali
2020 arXiv   pre-print
Two candidates for machine learning COVID-19 diagnosis are Computed Tomography (CT) scans and plain chest X-rays.  ...  Therefore, X-ray imagery paired with machine learning techniques is proposed a first-line triage tool for COVID-19 diagnostics.  ...  in Figure 6 represent True Positive, False Positive, True Negative, and False Negative percentages, respectively.  ... 
arXiv:2009.08831v1 fatcat:5jlg2lr7wzcpfbz223ilvj6pfi

Glioblastoma Synthesis and Segmentation with 3D Multi-Modal MRI: A Study using Generative Adversarial Networks

Edmond Wang
2021 International Journal on Computational Science & Applications  
The use of Deep Learning - in particular CNNs and GANs - have become prominent in dealing with various image segmentation and detection tasks.  ...  In this study, the history and various breakthroughs/challenges of utilising deep learning in glioblastoma detection is outlined and evaluated.  ...  This loss function was specifically designed to optimise segmentation on imbalanced medical datasets, being weighted by the constants 'alpha' and 'beta' which penalise false positive and false negatives  ... 
doi:10.5121/ijcsa.2021.11601 fatcat:4wjwcix2jrdrlckuoiwxnelhqq

AI'S Contribution to Ubiquitous Systems and Pervasive Networks Security – Reinforcement Learning vs Recurrent Networks

Christophe Feltus
2021 Journal of Ubiquitous Systems and Pervasive Networks  
Reinforcement learning and recurrent networks are two emerging machine-learning paradigms.  ...  The first learns the best actions an agent needs to perform to maximize its rewards in a particular environment and the second has the specificity to use an internal state to remember previous analysis  ...  [58] , who exploit LSTM in RNN units to generate an improved LSTM tree that has the ability of secondary detection to solve the problem of a high false negative in traditional RNN.  ... 
doi:10.5383/juspn.15.02.001 fatcat:tcfmazejvngihlmlqbt3gop72a
« Previous Showing results 1 — 15 out of 2,975 results