scholarly journals Scale-invariant localization using quasi-semantic object landmarks

2021 ◽  
Author(s):  
Andrew Holliday ◽  
Gregory Dudek
Keyword(s):  
State Of The Art ◽  
Visual Localization ◽  
Visual Feature ◽  
Scale Invariant ◽  
Object A ◽  
Change Factors ◽  
Major Scale ◽  
New Type ◽  
Semantic Object ◽  
Classical Point

AbstractThis work presents Object Landmarks, a new type of visual feature designed for visual localization over major changes in distance and scale. An Object Landmark consists of a bounding box $${\mathbf {b}}$$ b defining an object, a descriptor $${\mathbf {q}}$$ q of that object produced by a Convolutional Neural Network, and a set of classical point features within $${\mathbf {b}}$$ b . We evaluate Object Landmarks on visual odometry and place-recognition tasks, and compare them against several modern approaches. We find that Object Landmarks enable superior localization over major scale changes, reducing error by as much as 18% and increasing robustness to failure by as much as 80% versus the state-of-the-art. They allow localization under scale change factors up to 6, where state-of-the-art approaches break down at factors of 3 or more.

scholarly journals Robust CNN Compression Framework for Security-Sensitive Embedded Systems

2021 ◽  
Vol 11 (3) ◽  
pp. 1093
Author(s):  
Jeonghyun Lee ◽  
Sangkyun Lee
Keyword(s):  
Embedded Systems ◽  
Optimization Problem ◽  
State Of The Art ◽  
Classification Problems ◽  
Proximal Gradient Method ◽  
Knowledge Distillation ◽  
New Type ◽  
Adversarial Examples ◽  
Adversarial Training ◽  
Memory Efficient

Convolutional neural networks (CNNs) have achieved tremendous success in solving complex classification problems. Motivated by this success, there have been proposed various compression methods for downsizing the CNNs to deploy them on resource-constrained embedded systems. However, a new type of vulnerability of compressed CNNs known as the adversarial examples has been discovered recently, which is critical for security-sensitive systems because the adversarial examples can cause malfunction of CNNs and can be crafted easily in many cases. In this paper, we proposed a compression framework to produce compressed CNNs robust against such adversarial examples. To achieve the goal, our framework uses both pruning and knowledge distillation with adversarial training. We formulate our framework as an optimization problem and provide a solution algorithm based on the proximal gradient method, which is more memory-efficient than the popular ADMM-based compression approaches. In experiments, we show that our framework can improve the trade-off between adversarial robustness and compression rate compared to the existing state-of-the-art adversarial pruning approach.

scholarly journals Design and implementation of E-discovery as a service based on cloud computing

2013 ◽  
Vol 10 (2) ◽  
pp. 703-724 ◽  
Author(s):  
Taerim Lee ◽  
Hun Kim ◽  
Kyung-Hyune Rhee ◽  
Uk Shin
Keyword(s):  
Cloud Computing ◽  
Direct Contact ◽  
State Of The Art ◽  
Service Systems ◽  
Target System ◽  
Design And Implementation ◽  
Discovery Service ◽  
New Type ◽  
It Compliance ◽  
Network Device

Recently, as IT Compliance becomes more diverse, companies have to take a great amount of effort to comply with it and prepare countermeasures. Especially, E-Discovery is also one of the most notable compliances for IT and law. In order to minimize the time and cost for E-Discovery, many service systems and solutions using the state-of-the-art technology have been competitively developed. Among them, Cloud Computing is one of the most exclusive skills as a computing infrastructure for E-Discovery Service. Unfortunately, these products actually do not cover all kinds of E-Discovery works and have many drawbacks as well as considerable limitations. This paper, therefore, proposes a new type of E-Discovery Service Structure based on Cloud Computing called EDaaS(E-Discovery as a Service) to make the best usage of its advantages and overcome the limitations of the existing E-Discovery solutions. EDaaS enables E-Discovery participants to smoothly collaborate by removing constraints on working places and minimizing the number of direct contact with target systems. What those who want to use the EDaaS need is only a network device for using the Internet. Moreover, EDaaS can help to reduce the waste of time and human resources because no specific software to install on every target system is needed and the relatively exact time of completion can be obtained from it according to the amount of data for the manpower control. As a result of it, EDaaS can solve the litigant?s cost problem.

Orthonormal basis of wavelets with customizable frequency bands

2016 ◽  
Vol 14 (06) ◽  
pp. 1650050 ◽  
Author(s):  
Hiroshi Toda ◽  
Zhong Zhang
Keyword(s):  
Frequency Domain ◽  
Infinite Number ◽  
Orthonormal Basis ◽  
The Other ◽  
Just Intonation ◽  
Frequency Bands ◽  
Orthonormal Bases ◽  
Major Scale ◽  
Dilation Factor ◽  
New Type

We already proved the existence of an orthonormal basis of wavelets having an irrational dilation factor with an infinite number of wavelet shapes, and based on its theory, we proposed an orthonormal basis of wavelets with an arbitrary real dilation factor. In this paper, with the development of these fundamentals, we propose a new type of orthonormal basis of wavelets with customizable frequency bands. Its frequency bands can be freely designed with arbitrary bounds in the frequency domain. For example, we show two types of orthonormal bases of wavelets. One of them has an irrational dilation factor, and the other is designed based on the major scale in just intonation.

Causal Capture: Contextual Effects on the Perception of Collision Events

2002 ◽  
Vol 13 (6) ◽  
pp. 493-498 ◽  
Author(s):  
Brian J. Scholl ◽  
Ken Nakayama
Keyword(s):  
Contextual Effects ◽  
Contextual Effect ◽  
Stationary Object ◽  
Object A ◽  
Visual Events ◽  
New Type ◽  
Test Event ◽  
High Level ◽  
The Impact ◽  
Launch Event

In addition to perceiving the colors, shapes, and motions of objects, observers can perceive higher-level properties of visual events. One such property is causation, as when an observer sees one object cause another object to move by colliding with it. We report a striking new type of contextual effect on the perception of such collision events. Consider an object (A) that moves toward a stationary object (B) until they are adjacent, at which point A stops and B starts moving along the same path. Such “launches” are perceived in terms beyond these kinematics: As noted in Michotte's classic studies, observers perceive A as being the cause of B's motion. When A and B fully overlap before B's motion, however, observers often see this test event as a completely noncausal “pass”: One object remains stationary while another passes over it. When a distinct launch event occurs nearby, however, the test event is “captured”: It too is now irresistibly seen as causal. For this causal capture to occur, the context event need be present for only 50 ms surrounding the “impact,” but capture is destroyed by only 200 ms of temporal asynchrony between the two events. We report a study of such cases, and others, that help define the rules that the visual system uses to construct percepts of seemingly high-level properties like causation.

scholarly journals Zero-Shot Learning from Adversarial Feature Residual to Compact Visual Feature

2020 ◽  
Vol 34 (07) ◽  
pp. 11547-11554
Author(s):  
Bo Liu ◽  
Qiulei Dong ◽  
Zhanyi Hu
Keyword(s):  
State Of The Art ◽  
Feature Space ◽  
Visual Features ◽  
Selection Strategy ◽  
Semantic Features ◽  
Visual Feature ◽  
Adversarial Network ◽  
Benchmark Datasets ◽  
Residual Generator ◽  
Object Features

Recently, many zero-shot learning (ZSL) methods focused on learning discriminative object features in an embedding feature space, however, the distributions of the unseen-class features learned by these methods are prone to be partly overlapped, resulting in inaccurate object recognition. Addressing this problem, we propose a novel adversarial network to synthesize compact semantic visual features for ZSL, consisting of a residual generator, a prototype predictor, and a discriminator. The residual generator is to generate the visual feature residual, which is integrated with a visual prototype predicted via the prototype predictor for synthesizing the visual feature. The discriminator is to distinguish the synthetic visual features from the real ones extracted from an existing categorization CNN. Since the generated residuals are generally numerically much smaller than the distances among all the prototypes, the distributions of the unseen-class features synthesized by the proposed network are less overlapped. In addition, considering that the visual features from categorization CNNs are generally inconsistent with their semantic features, a simple feature selection strategy is introduced for extracting more compact semantic visual features. Extensive experimental results on six benchmark datasets demonstrate that our method could achieve a significantly better performance than existing state-of-the-art methods by ∼1.2-13.2% in most cases.

THE FIXED SCALE TRANSFORMATION: STATUS AND PERSPECTIVES

Fractals ◽  
1993 ◽  
Vol 01 (03) ◽  
pp. 650-662 ◽  
Author(s):  
L. PIETRONERO
Keyword(s):  
Growth Models ◽  
Path Integrals ◽  
Coarse Grained ◽  
Scale Transformation ◽  
Scale Invariant ◽  
Simplified Models ◽  
Self Organized ◽  
Self Organized Criticality ◽  
New Type ◽  
Pile Model

Irreversible fractal growth models like DLA and DBM have confronted us with theoretical problems of a new type that cannot be described in terms of the standard concepts like field theory and the renormalization group. The Fixed Scale Transformation is a theoretical scheme of a new type that is able to treat these problems in a reasonably systematic way. The idea is to focus on the dynamics at a given scale and to compute accurately the correlations at this scale by suitable lattice path integrals. The use of scale invariant growth rules then allows the generalization of these correlations to coarse-grained cells of any size and therefore to obtain the fractal dimension. We summarize the present status of the FST approach by focusing on the most recent results about the scale invariant dynamics of DLA/DBM. The possible extensions to other problems like the sand pile model (self-organized-criticality) and simplified models of turbulence will also be considered.

scholarly journals A Visual SLAM Robust against Dynamic Objects Based on Hybrid Semantic-Geometry Information

2021 ◽  
Vol 10 (10) ◽  
pp. 673
Author(s):  
Sheng Miao ◽  
Xiaoxiong Liu ◽  
Dazheng Wei ◽  
Changze Li
Keyword(s):  
Moving Objects ◽  
State Of The Art ◽  
Semantic Segmentation ◽  
Visual Localization ◽  
Epipolar Constraint ◽  
Trajectory Error ◽  
Localization And Mapping ◽  
Dynamic Objects ◽  
Public Datasets ◽  
Localization Approach

A visual localization approach for dynamic objects based on hybrid semantic-geometry information is presented. Due to the interference of moving objects in the real environment, the traditional simultaneous localization and mapping (SLAM) system can be corrupted. To address this problem, we propose a method for static/dynamic image segmentation that leverages semantic and geometric modules, including optical flow residual clustering, epipolar constraint checks, semantic segmentation, and outlier elimination. We integrated the proposed approach into the state-of-the-art ORB-SLAM2 and evaluated its performance on both public datasets and a quadcopter platform. Experimental results demonstrated that the root-mean-square error of the absolute trajectory error improved, on average, by 93.63% in highly dynamic benchmarks when compared with ORB-SLAM2. Thus, the proposed method can improve the performance of state-of-the-art SLAM systems in challenging scenarios.

The state-of-the-art coronary stent with crystallized sirolimus: the MiStent technology and its clinical program

2020 ◽  
Author(s):  
Masafumi Ono ◽  
Kuniaki Takahashi ◽  
Chao Gao ◽  
Hideyuki Kawashima ◽  
Xinlei Wu ◽  
...  
Keyword(s):  
State Of The Art ◽  
Standard Of Care ◽  
Controlled Drug Release ◽  
Coronary Intervention ◽  
Clinical Program ◽  
Drug Eluting ◽  
New Type ◽  
Artery Disease ◽  
And Performance ◽  
Sirolimus Eluting Stent

Drug-eluting stents (DES) have been developed over recent decades and the implantation of DES is the standard of care in contemporary percutaneous coronary intervention for patients with coronary artery disease. The MiStent sirolimus-eluting stent has several unique features; ultra-thin (64 μm) struts, a bioresorbable polymer and a controlled drug release from microcrystalline sirolimus as a reservoir embedded in the vessel wall. Results of recent clinical trials demonstrated the potential performance of this state-of-the-art DES. In the present review, we provide an overview of the development of DES, in particular the design and performance of the novel MiStent sirolimus-eluting stent from technological and clinical points of view and discuss the potentials of this new type of DES.

scholarly journals Towards Cognitive Recommender Systems

Algorithms ◽  
2020 ◽  
Vol 13 (8) ◽  
pp. 176
Author(s):  
Amin Beheshti ◽  
Shahpar Yakhchi ◽  
Salman Mousaeirad ◽  
Seyed Mohssen Ghafari ◽  
Srinivasa Reddy Goluguri ◽  
...  
Keyword(s):  
Recommender Systems ◽  
State Of The Art ◽  
User Preferences ◽  
Data Driven ◽  
New Approach ◽  
Domain Experts ◽  
Adaptive Mechanisms ◽  
New Type ◽  
Cognitive Activities ◽  
Time Aware

Intelligence is the ability to learn from experience and use domain experts’ knowledge to adapt to new situations. In this context, an intelligent Recommender System should be able to learn from domain experts’ knowledge and experience, as it is vital to know the domain that the items will be recommended. Traditionally, Recommender Systems have been recognized as playlist generators for video/music services (e.g., Netflix and Spotify), e-commerce product recommenders (e.g., Amazon and eBay), or social content recommenders (e.g., Facebook and Twitter). However, Recommender Systems in modern enterprises are highly data-/knowledge-driven and may rely on users’ cognitive aspects such as personality, behavior, and attitude. In this paper, we survey and summarize previously published studies on Recommender Systems to help readers understand our method’s contributions to the field in this context. We discuss the current limitations of the state of the art approaches in Recommender Systems and the need for our new approach: A vision and a general framework for a new type of data-driven, knowledge-driven, and cognition-driven Recommender Systems, namely, Cognitive Recommender Systems. Cognitive Recommender Systems will be the new type of intelligent Recommender Systems that understand the user’s preferences, detect changes in user preferences over time, predict user’s unknown favorites, and explore adaptive mechanisms to enable intelligent actions within the compound and changing environments. We present a motivating scenario in banking and argue that existing Recommender Systems: (i) do not use domain experts’ knowledge to adapt to new situations; (ii) may not be able to predict the ratings or preferences a customer would give to a product (e.g., loan, deposit, or trust service); and (iii) do not support data capture and analytics around customers’ cognitive activities and use it to provide intelligent and time-aware recommendations.

Sign in / Sign up

Export Citation Format

Share Document