4,738 Hits in 3.9 sec

DOOM Level Generation using Generative Adversarial Networks [article]

Edoardo Giacomello and Pier Luca Lanzi and Daniele Loiacono
2018 arXiv   pre-print
We applied Generative Adversarial Networks (GANs) to learn a model of DOOM levels from human-designed content. Initially, we analysed the levels and extracted several topological features.  ...  We used the two networks to generate new levels and compared the results to assess whether the network trained using also the topological features could generate levels more similar to human-designed ones  ...  DEEP LEVEL GENERATION In this study, we applied Generative Adversarial Networks [34] (GANs) to learn a model of existing DOOM levels; we then used the model to generate new levels.  ... 
arXiv:1804.09154v1 fatcat:obshz7smxnarlnbl3eplzwydti


Mohit Sewak, Sanjay K. Sahay, Hemant Rathore
2020 Adjunct Proceedings of the 2020 ACM International Joint Conference on Pervasive and Ubiquitous Computing and Proceedings of the 2020 ACM International Symposium on Wearable Computers  
We designed and developed DOOM (Adversarial-DRL based Opcode level Obfuscator to generate Metamorphic malware), a novel system that uses adversarial deep reinforcement learning to obfuscate malware at  ...  To the best of our knowledge, DOOM is the first system that could generate obfuscated malware detailed to individual op-code level.  ...  There has been attempt to use Convolutional Neural Networks (CNN) based Generative Adversarial Networks (GANs) [16] , [6] , [7] , [15] as well.  ... 
doi:10.1145/3410530.3414411 dblp:conf/huc/SewakSR20 fatcat:7kplgi5wcbht5gvu2ukyrhu4wq

Almost-Everywhere Secure Computation with Edge Corruptions

Nishanth Chandran, Juan A. Garay, Rafail Ostrovsky
2013 Journal of Cryptology  
What makes the problem more challenging is to consider the case of sparse networks.  ...  Note that if an adversary corrupts an edge, even if the two nodes that share that edge are honest, the adversary can control the link and thus deliver wrong messages to both players.  ...  Another interesting question is whether we can obtain polynomial-time protocols for a.e. secure computation when a constant fraction of the edges are corrupted on constant-degree networks.  ... 
doi:10.1007/s00145-013-9176-3 fatcat:fekumis55rgetj6wnuzuka2cfe

Doom as an interface for process management

Dennis Chao
2001 Proceedings of the SIGCHI conference on Human factors in computing systems - CHI '01  
Instead of creating an interface de novo for the task, the author modified a popular computer game, Doom, to perform useful work.  ...  In addition, the application highlights the adversarial relationships that exist in a computer and suggests a new resource allocation scheme.  ...  People with different levels of authority can be given weapons of different strengths.  ... 
doi:10.1145/365024.365078 dblp:conf/chi/Chao01 fatcat:2d7cm2t4erbifaw526kmtxjpxe

Experience enrichment based task independent reward model [article]

Min Xu
2017 arXiv   pre-print
In this paper, we propose an implicit generic reward model for reinforcement learning. Unlike those rewards that are manually defined for specific tasks, such implicit reward is task independent.  ...  To model the enrichment of experiences, we first learn a low dimension representation of S using unsupervised learning, such as Generative Adversarial Networks (GAN) [2] .  ...  Unsupervised deep learning techniques such as Generative Adversarial Networks (GAN) [2] has been shown to be able to capture the intrinsic distribution on the manifold among images.  ... 
arXiv:1705.07460v1 fatcat:hr2torqic5ck3juickqzgng5ra

Leveraging Extracted Model Adversaries for Improved Black Box Attacks [article]

Naveen Jafer Nizar, Ari Kobren
2020 arXiv   pre-print
Second, we use our own white box method to generate input perturbations that cause the approximate model to fail. These perturbed inputs are used against the victim.  ...  We present a method for adversarial input generation against black box models for reading comprehension based question answering. Our approach is composed of two steps.  ...  networks?  ... 
arXiv:2010.16336v2 fatcat:wptgq2r7vvfh7fsuqujwq2lzx4

Edge Fault Tolerance on Sparse Networks [chapter]

Nishanth Chandran, Juan Garay, Rafail Ostrovsky
2012 Lecture Notes in Computer Science  
We remark that allowing an adversary to corrupt edges in the network can be seen as taking a step closer towards guaranteeing a.e. agreement amongst honest nodes even on adversarially chosen communication  ...  The number of such nodes is a function of the underlying communication graph and the adversarial set of nodes.  ...  In an ideal scenario, we would like to construct a.e. computation protocols on arbitrary adversarially chosen communication networks. Unfortunately, this is impossible in general 1 .  ... 
doi:10.1007/978-3-642-31585-5_41 fatcat:owpxxqqg4jgx3g6zqlzg2faaey

Level generation and style enhancement – deep learning for game development overview [article]

Piotr Migdał, Bartłomiej Olechno, Błażej Podgórski
2021 arXiv   pre-print
In particular, we include: - Generative Adversarial Networks for creating new images from existing examples (e.g.  ...  We present seven approaches to create level maps, each using statistical methods, machine learning, or deep learning.  ...  This task involves a few levels of complexity -from the high-level general pattern of the map to low-level details and decorations such as tire tracks.  ... 
arXiv:2107.07397v1 fatcat:usno4wbir5gsjdv44cy5kikzka

Bootstrapping Conditional GANs for Video Game Level Generation [article]

Ruben Rodriguez Torrado, Ahmed Khalifa, Michael Cerny Green, Niels Justesen, Sebastian Risi, Julian Togelius
2019 arXiv   pre-print
Generative Adversarial Networks (GANs) have shown im-pressive results for image generation.  ...  In this paper, we propose a new GAN architec-ture namedConditional Embedding Self-Attention Genera-tive Adversarial Network(CESAGAN) and a new bootstrap-ping training procedure.  ...  All authors acknowledge Per Josefsen and Nicola Zaltron, who were responsible for the 45 human-designed levels.  ... 
arXiv:1910.01603v1 fatcat:6uazxl2n3ncv5imokuflvqf73e

Game Sprite Generator Using a Multi Discriminator GAN

2019 KSII Transactions on Internet and Information Systems  
This paper proposes an image generation method using a Multi Discriminator Generative Adversarial Net (MDGAN) as a next generation 2D game sprite creation technique.  ...  The experimental results demonstrate that our MDGAN can be used for 2D image sprite generation and modification work with little manual cost.  ...  Their results showed that GANs could generate structure of DOOM levels in first person shooter games [34] .  ... 
doi:10.3837/tiis.2019.08.025 fatcat:jpergpgaxncjrgb4lwnnchlz4a

Learning to Act by Predicting the Future [article]

Alexey Dosovitskiy, Vladlen Koltun
2017 arXiv   pre-print
A model trained using the presented approach won the Full Deathmatch track of the Visual Doom AI Competition, which was held in previously unseen environments.  ...  The results also show that trained models successfully generalize across environments and goals.  ...  In a three-dimensional computer game, they can include health, ammunition levels, and the number of adversaries overcome.  ... 
arXiv:1611.01779v2 fatcat:aone2xwqlrfqva3vtljm5kjhoe

Task-Relevant Object Discovery and Categorization for Playing First-person Shooter Games [article]

Junchi Liang, Abdeslam Boularias
2018 arXiv   pre-print
Experiments on the game Doom provide a good evidence for the benefit of this approach.  ...  The result is encoded as a vector indicating objects that are in the frame and their locations, and used as a side input to DRQN.  ...  Along these lines, Dosovitskiy and Koltun 2016 modified the DQN structure by adding a network that takes as inputs low-dimensional sensory data, such as health, ammunition levels, and the number of adversaries  ... 
arXiv:1806.06392v1 fatcat:sph6lhakprde7ovrcjqbjbj5ky

Security issues and defensive approaches in deep learning frameworks

Hongsong Chen, Yongpeng Zhang, Yongrui Cao, Jing Xie
2021 Tsinghua Science and Technology  
Moreover, we analyze a case of the physical-world use of deep learning security issues. In addition, we discuss future directions and open issues in deep learning frameworks.  ...  [8] 2019 RMSG Proposed an adversarial method generating perturbations based on root mean square gradient, which formulates the adversarial perturbation size in the root mean square level and updates  ...  The FGSM and DeepFool [16] are methods for generating adversarial samples, and both are white-box attacks. In a neural network, back propagation is used to minimize the loss function.  ... 
doi:10.26599/tst.2020.9010050 fatcat:zeklghvvurhdzpozagqs32g67a

LSGAN-AT: enhancing malware detector robustness against adversarial examples

Jianhua Wang, Xiaolin Chang, Yixiang Wang, Ricardo J. Rodríguez, Jianan Zhang
2021 Cybersecurity  
AT module makes adversarial training using AME generated by LSGAN to generate ML-based Robust Malware Detector (RMD).  ...  Generative Adversarial Network (GAN) is a kind of AME generation method, but the existing GAN-based AME generation methods have the issues of inadequate optimization, mode collapse and training instability  ...  After that, we utilize Least Square (LS) loss in the Generative Adversarial Network (GAN) in terms of a generator and a discriminator with brand-new network structures for adversarial training to generate  ... 
doi:10.1186/s42400-021-00102-9 fatcat:udahc3q76fbexpciuzersfcpcy

Evolving Mario Levels in the Latent Space of a Deep Convolutional Generative Adversarial Network [article]

Vanessa Volz, Jacob Schrum, Jialin Liu, Simon M. Lucas, Adam Smith, Sebastian Risi
2018 arXiv   pre-print
Generative Adversarial Networks (GANs) are a machine learning approach capable of generating novel example outputs across a space of provided training examples.  ...  This paper trains a GAN to generate levels for Super Mario Bros using a level from the Video Game Level Corpus.  ...  Generative Adversarial Networks Generative Adversarial Networks (GANs) were first introduced by Goodfellow et al. [7] in 2014.  ... 
arXiv:1805.00728v1 fatcat:xwabyb2drjcnjjbhtpohz3oniy
« Previous Showing results 1 — 15 out of 4,738 results