Filters








146,148 Hits in 8.4 sec

InDiD: Instant Disorder Detection via Representation Learning [article]

Evgenia Romanenkova and Alexander Stepikin and Matvey Morozov and Alexey Zaytsev
2022 arXiv   pre-print
We propose a principled loss function that balances change detection delay and time to a false alarm.  ...  Classic approaches for change point detection (CPD) might underperform for semi-structured sequential data because they cannot process its structure without a proper representation.  ...  In our work, we tried to bridge this gap and proposed a principled solution for representation learning devoted to a change point detection problem.  ... 
arXiv:2106.02602v3 fatcat:zor2uyfbbrglnjru7ckaehfk2y

Learning a Metacognition for Object Detection [article]

Marlene Berke, Mario Belledonne, Zhangir Azerbayev, Julian Jara-Ettinger
2022 arXiv   pre-print
Given noisy output from an object-detection model, METAGEN learns a meta-representation of how its perceptual system works and uses it to infer the objects in the world responsible for the detections.  ...  We find that METAGEN quickly learns an accurate metacognitive representation of the neural network, and that this improves detection accuracy by filling in objects that the detection model missed and removing  ...  Conditioning inference on these basic principles enables METAGEN to learn a metacognitive representation of the object detection model by analyzing and resolving patterns of percepts that violate the Spelke  ... 
arXiv:2110.03105v2 fatcat:gwmsgxlwhfb6zoxmea6v3x6754

A survey of appearance models in visual object tracking

Xi Li, Weiming Hu, Chunhua Shen, Zhongfei Zhang, Anthony Dick, Anton Van Den Hengel
2013 ACM Transactions on Intelligent Systems and Technology  
In particular, this survey takes a module-based architecture that enables readers to easily grasp the key points of visual object tracking.  ...  To help readers swiftly learn the recent advances in 2D appearance models for visual object tracking, we contribute this survey, which provides a detailed review of the existing 2D appearance models.  ...  HGDAMs via multi-layer combination. In principle, the goal of the HGDAMs via multi-layer combination is to combine the information from the generative and discriminative models at multiple layers.  ... 
doi:10.1145/2508037.2508039 fatcat:uwptu4nkmbhjvib5szoafjhg3i

A Survey of Appearance Models in Visual Object Tracking [article]

Xi Li, Weiming Hu, Chunhua Shen, Zhongfei Zhang, Anthony Dick, Anton van den Hengel
2013 arXiv   pre-print
In particular, this survey takes a module-based architecture that enables readers to easily grasp the key points of visual object tracking.  ...  To help readers swiftly learn the recent advances in 2D appearance models for visual object tracking, we contribute this survey, which provides a detailed review of the existing 2D appearance models.  ...  HGDAMs via multi-layer combination. In principle, the goal of the HGDAMs via multi-layer combination is to combine the information from the generative and discriminative models at multiple layers.  ... 
arXiv:1303.4803v1 fatcat:tx333ej63faufnxffjttac4jxq

Enhancing Robustness of Deep Neural Networks Against Adversarial Malware Samples: Principles, Framework, and AICS'2019 Challenge [article]

Deqiang Li, Qianmu Li, Yanfang Ye, Shouhuai Xu
2020 arXiv   pre-print
This naturally calls for machine learning based malware detection.  ...  Some of these principles have been scattered in the literature, but others are proposed in this paper for the first time.  ...  Both the principles and framework should be seen as a starting point and systematically refined in the future.  ... 
arXiv:1812.08108v3 fatcat:4trysg2ipnfj7bspyblvtan2eq

POSE-VIWEPOINT ADAPTIVE OBJECT TRACKING VIA ONLINE LEARNING APPROACH

Vinayagam Mariappan, Hyung-O Kim, Minwoo Lee, Juphil Cho, Jaesang Cha
2015 International journal of advanced smart convergence  
The data-dependent adaptive appearance models often encounter the drift problems because the online algorithms does not get the required amount of data for online learning.  ...  In this paper, we propose an effective tracking algorithm with an appearance model based on features extracted from a video frame with posture variation and camera view point adaptation by employing the  ...  ONLINE LEARNING TLD (Tracking-Learning-Detection) [1] algorithm is a famous online learning tracking algorithm.  ... 
doi:10.7236/ijasc.2015.4.2.20 fatcat:nskcspbob5djvky5d4h37i3lwa

A Framework for Enhancing Deep Neural Networks Against Adversarial Malware [article]

Deqiang Li, Qianmu Li, Yanfang Ye, Shouhuai Xu
2021 arXiv   pre-print
Machine learning-based malware detection is known to be vulnerable to adversarial evasion attacks. The state-of-the-art is that there are no effective defenses against these attacks.  ...  Some of these principles have been scattered in the literature, but the others are introduced in this paper for the first time.  ...  /regularization, and DAEbased representation learning.  ... 
arXiv:2004.07919v2 fatcat:ug4ix4sbdjdkjfniqaabdchaga

Anomaly Detection & Behavior Prediction: Higher-Level Fusion Based on Computational Neuroscientific Principles [chapter]

Bradley J., Neil A., Majid Zandipour, Lauren H., Denis Garagic, James R., Michael Seibert
2009 Sensor and Data Fusion  
Pre-defined activity patterns can be detected and identified to operators.  ...  It is impractical to consider a rule-based approach for achieving such a task, so an adaptive method is required: that is, a capability to learn what is normal in a scene is required.  ...  Weights between grid locations change via presynaptically gated Hebbian learning.  ... 
doi:10.5772/6585 fatcat:2yks6ecbovh3xkw5c5uzdcneo4

Towards a model for tool-body assimilation and adaptive tool-use

Cota Nabeshima, Yasuo Kuniyoshi, Max Lungarella
2007 2007 IEEE 6th International Conference on Development and Learning  
This form of perceptual learning occurs at all stages of life.  ...  We evaluate our model by instantiating it in a simulated tool-using robot which learns to handle tools of various shapes to retrieve an object placed out of sight and out of reach.  ...  Such learning of motion change by contact is essentially a problem of sensory integration.  ... 
doi:10.1109/devlrn.2007.4354031 fatcat:3aux3u53ivbrnbk7zj2pn4bndi

Self-taught learning of a deep invariant representation for visual tracking via temporal slowness principle

Jason Kuen, Kian Ming Lim, Chin Poo Lee
2015 Pattern Recognition  
In this paper, we propose to learn complex-valued invariant representations from tracked sequential image patches, via strong temporal slowness constraint and stacked convolutional autoencoders.  ...  With the learned representation and online training samples, a logistic regression classifier is adopted to distinguish target from background, and retrained online to adapt to appearance changes.  ...  For interest point detection, two approaches are first considered.  ... 
doi:10.1016/j.patcog.2015.02.012 fatcat:26lq2q5uvnduxcdtiqttm3cgui

Spatio-temporal Self-Supervised Representation Learning for 3D Point Clouds [article]

Siyuan Huang, Yichen Xie, Song-Chun Zhu, Yixin Zhu
2021 arXiv   pre-print
Moreover, the spatio-temporal contextual cues embedded in 3D point clouds significantly improve the learned representations.  ...  In this paper, we tackle this challenge by introducing a spatio-temporal representation learning (STRL) framework, capable of learning from unlabeled 3D point clouds in a self-supervised fashion.  ...  To adhere to the above principles and tackle the challenges introduced thereby, we devise a spatio-temporal representation learning (STRL) framework to learn from unlabeled 3D point clouds.  ... 
arXiv:2109.00179v1 fatcat:pv7hktdu5bdfhootjdca3rheey

Enhancing an instructional design model for virtual reality-based learning

Chwen Jen Chen, Chee Siong Teh
2013 Australasian Journal of Educational Technology  
learning environments.  ...  <p>In order to effectively utilize the capabilities of virtual reality (VR) in supporting the desired learning outcomes, careful consideration in the design of instruction for VR learning is crucial.  ...  The learning environment sometimes provided coaching via pop-up text messages with narration when it detected the learner breaking a traffic rule.  ... 
doi:10.14742/ajet.247 fatcat:nxczjikekjcwtmv3enp3jlztdu

Complex Declarative Learning [chapter]

Linda Bol, Douglas J. Hacker, Andrew Mattarella-Micke, Sian L. Beilock, Norbert M. Seel, Claus Andreas Foss Rosenstand, Lowell Dean Tong, Christian Burke, Ann N. Poncelet, Shawn Ell, Monica Zilioli, Fabian A. Soto (+190 others)
2012 Encyclopedia of the Sciences of Learning  
It embodies concepts, principles, ideas, schemas, and theories (Ohlsson, 1994; .  ...  existing knowledge, retrieving appropriate analogies, producing explanations, coordinating different representations and perspectives, abandoning or rejecting prior concepts that are no longer useful,  ...  , and why many of the learning mechanisms cannot in principle produce true nonmonotonic learning.  ... 
doi:10.1007/978-1-4419-1428-6_295 fatcat:hr2ly5ok6nadfc7c5wyfw4zlom

The Vision of Self-Evolving Computing Systems [article]

Danny Weyns, Thomas Baeck, Rene Vidal, Xin Yao, Ahmed Nabil Belbachir
2022 arXiv   pre-print
Specifically, when a self-evolving computing system detects conditions outside its operational domain, such as an anomaly or a new goal, it activates an evolutionary engine that runs online experiments  ...  A key aspect of this sustainability is the ability of computing systems to cope with the continuous change they face, ranging from dynamic operating conditions, to changing goals, and technological progress  ...  Evolutionary Learning Engine Computing Warehouse Detection Evolutionary Learning Pipeline Change Enactment Anomaly Detection New Goal Detection Warehouse Manager Catalog Sandbox Historical Data Adaptive  ... 
arXiv:2204.06825v1 fatcat:con5dhmfkbaahavua4utkfnsfy

Learning Taxonomies of Concepts and not Words using Contextualized Word Representations: A Position Paper [article]

Lukas Schmelzeisen, Steffen Staab
2019 arXiv   pre-print
We outline a novel approach for taxonomy learning that (1) defines concepts as synsets, (2) learns density-based approximations of contextualized word representations, and (3) can measure similarity and  ...  One limitation of current taxonomy learning systems is that they define concepts as single words.  ...  Discussion To the best of our knowledge, there is no taxonomy learning system providing a principled way to model multiple word senses.  ... 
arXiv:1902.02169v1 fatcat:xehjdyjdnfb23fxiyf7wlyw4wq
« Previous Showing results 1 — 15 out of 146,148 results