IA Scholar Query: Point Cloud Denoising via Momentum Ascent in Gradient Fields.
https://scholar.archive.org/
Internet Archive Scholar query results feedeninfo@archive.orgWed, 20 Jul 2022 00:00:00 GMTfatcat-scholarhttps://scholar.archive.org/help1440Exploring the Devil in Graph Spectral Domain for 3D Point Cloud Attacks
https://scholar.archive.org/work/gsl5ultqbfbgnh5lu5muaav76y
3D dynamic point clouds provide a discrete representation of real-world objects or scenes in motion, which have been widely applied in immersive telepresence, autonomous driving, surveillance, etc. However, point clouds acquired from sensors are usually perturbed by noise, which affects downstream tasks such as surface reconstruction and analysis. Although many efforts have been made for static point cloud denoising, few works address dynamic point cloud denoising. In this paper, we propose a novel gradient-based dynamic point cloud denoising method, exploiting the temporal correspondence for the estimation of gradient fields – also a fundamental problem in dynamic point cloud processing and analysis. The gradient field is the gradient of the log-probability function of the noisy point cloud, based on which we perform gradient ascent so as to converge each point to the underlying clean surface. We estimate the gradient of each surface patch by exploiting the temporal correspondence, where the temporally corresponding patches are searched leveraging on rigid motion in classical mechanics. In particular, we treat each patch as a rigid object, which moves in the gradient field of an adjacent frame via force until reaching a balanced state, i.e., when the sum of gradients over the patch reaches 0. Since the gradient would be smaller when the point is closer to the underlying surface, the balanced patch would fit the underlying surface well, thus leading to the temporal correspondence. Finally, the position of each point in the patch is updated along the direction of the gradient averaged from corresponding patches in adjacent frames. Experimental results demonstrate that the proposed model outperforms state-of-the-art methods.Qianjiang Hu, Daizong Liu, Wei Huwork_gsl5ultqbfbgnh5lu5muaav76yWed, 20 Jul 2022 00:00:00 GMTModern applications of machine learning in quantum sciences
https://scholar.archive.org/work/rhehow62v5ci7hzdraj4ypj5ki
In these Lecture Notes, we provide a comprehensive introduction to the most recent advances in the application of machine learning methods in quantum sciences. We cover the use of deep learning and kernel methods in supervised, unsupervised, and reinforcement learning algorithms for phase classification, representation of many-body quantum states, quantum feedback control, and quantum circuits optimization. Moreover, we introduce and discuss more specialized topics such as differentiable programming, generative models, statistical approach to machine learning, and quantum machine learning.Anna Dawid, Julian Arnold, Borja Requena, Alexander Gresch, Marcin Płodzień, Kaelan Donatella, Kim A. Nicoli, Paolo Stornati, Rouven Koch, Miriam Büttner, Robert Okuła, Gorka Muñoz-Gil, Rodrigo A. Vargas-Hernández, Alba Cervera-Lierta, Juan Carrasquilla, Vedran Dunjko, Marylou Gabrié, Patrick Huembeli, Evert van Nieuwenburg, Filippo Vicentini, Lei Wang, Sebastian J. Wetzel, Giuseppe Carleo, Eliška Greplová, Roman Krems, Florian Marquardt, Michał Tomza, Maciej Lewenstein, Alexandre Dauphinwork_rhehow62v5ci7hzdraj4ypj5kiFri, 24 Jun 2022 00:00:00 GMTSelected Inductive Biases in Neural Networks To Generalize Beyond the Training Domain
https://scholar.archive.org/work/4wtxz5voxrgqtjdht7jbmpd63u
Die künstlichen neuronalen Netze des computergesteuerten Sehens können mit den vielf\"altigen Fähigkeiten des menschlichen Sehens noch lange nicht mithalten. Im Gegensatz zum Menschen können künstliche neuronale Netze durch kaum wahrnehmbare Störungen durcheinandergebracht werden, es mangelt ihnen an Generalisierungsfähigkeiten über ihre Trainingsdaten hinaus und sie benötigen meist noch enorme Datenmengen für das Erlernen neuer Aufgaben. Somit sind auf neuronalen Netzen basierende Anwendungen häufig auf kleine Bereiche oder kontrollierte Umgebungen beschränkt und lassen sich schlecht auf andere Aufgaben übertragen. In dieser Dissertation, werden vier Veröffentlichungen besprochen, die sich mit diesen Einschränkungen auseinandersetzen und Algorithmen im Bereich des visuellen Repräsentationslernens weiterentwickeln. In der ersten Veröffentlichung befassen wir uns mit dem Erlernen der unabhängigen Faktoren, die zum Beispiel eine Szenerie beschreiben. Im Gegensatz zu vorherigen Arbeiten in diesem Forschungsfeld verwenden wir hierbei jedoch weniger künstliche, sondern natürlichere Datensätze. Dabei beobachten wir, dass die zeitlichen Änderungen von Szenerien beschreibenden, natürlichen Faktoren (z.B. die Positionen von Personen in einer Fußgängerzone) einer verallgemeinerten Laplace-Verteilung folgen. Wir nutzen die verallgemeinerte Laplace-Verteilung als schwaches Lernsignal, um neuronale Netze für mathematisch beweisbares Repräsentationslernen unabhängiger Faktoren zu trainieren. Wir erzielen in den disentanglement_lib Wettbewerbsdatensätzen vergleichbare oder bessere Ergebnisse als vorherige Arbeiten – dies gilt auch für die von uns beigesteuerten Datensätze, welche natürliche Faktoren beinhalten. Die zweite Veröffentlichung untersucht, ob verschiedene neuronale Netze bereits beobachtete, eine Szenerie beschreibende Faktoren generalisieren können. In den meisten bisherigen Generalisierungswettbewerben werden erst während der Testphase neue Störungsfaktoren hinzugefügt - wir hingegen garantieren, dass die für die T [...]Lukas Schott, Universitaet Tuebingen, Bethge, Matthias (Prof. Dr.)work_4wtxz5voxrgqtjdht7jbmpd63uTue, 12 Apr 2022 00:00:00 GMTPoint Cloud Denoising via Momentum Ascent in Gradient Fields
https://scholar.archive.org/work/w3jhw35yhfd5phxml7vk34ifxu
To achieve point cloud denoising, traditional methods heavily rely on geometric priors, and most learning-based approaches suffer from outliers and loss of details. Recently, the gradient-based method was proposed to estimate the gradient fields from the noisy point clouds using neural networks, and refine the position of each point according to the estimated gradient. However, the predicted gradient could fluctuate, leading to perturbed and unstable solutions, as well as a large inference time. To address these issues, we develop the momentum gradient ascent method that leverages the information of previous iterations in determining the trajectories of the points, thus improving the stability of the solution and reducing the inference time. Experiments demonstrate that the proposed method outperforms state-of-the-art methods with a variety of point clouds and noise levels.Yaping Zhao, Haitian Zheng, Zhongrui Wang, Jiebo Luo, Edmund Y. Lamwork_w3jhw35yhfd5phxml7vk34ifxuTue, 15 Mar 2022 00:00:00 GMTCrystal Diffusion Variational Autoencoder for Periodic Material Generation
https://scholar.archive.org/work/5dbnwezv4raqvdthysxss65cl4
Generating the periodic structure of stable materials is a long-standing challenge for the material design community. This task is difficult because stable materials only exist in a low-dimensional subspace of all possible periodic arrangements of atoms: 1) the coordinates must lie in the local energy minimum defined by quantum mechanics, and 2) global stability also requires the structure to follow the complex, yet specific bonding preferences between different atom types. Existing methods fail to incorporate these factors and often lack proper invariances. We propose a Crystal Diffusion Variational Autoencoder (CDVAE) that captures the physical inductive bias of material stability. By learning from the data distribution of stable materials, the decoder generates materials in a diffusion process that moves atomic coordinates towards a lower energy state and updates atom types to satisfy bonding preferences between neighbors. Our model also explicitly encodes interactions across periodic boundaries and respects permutation, translation, rotation, and periodic invariances. We significantly outperform past methods in three tasks: 1) reconstructing the input structure, 2) generating valid, diverse, and realistic materials, and 3) generating materials that optimize a specific property. We also provide several standard datasets and evaluation metrics for the broader machine learning community.Tian Xie, Xiang Fu, Octavian-Eugen Ganea, Regina Barzilay, Tommi Jaakkolawork_5dbnwezv4raqvdthysxss65cl4Mon, 14 Mar 2022 00:00:00 GMTRobust, Deep, and Reinforcement Learning for Management of Communication and Power Networks
https://scholar.archive.org/work/ltekzvvpuncclag5vmbj2qrhau
This thesis develops data-driven machine learning algorithms to managing and optimizing the next-generation highly complex cyberphysical systems, which desperately need ground-breaking control, monitoring, and decision making schemes that can guarantee robustness, scalability, and situational awareness. The present thesis first develops principled methods to make generic machine learning models robust against distributional uncertainties and adversarial data. Particular focus will be on parametric models where some training data are being used to learn a parametric model. The developed framework is of high interest especially when training and testing data are drawn from "slightly" different distribution. We then introduce distributionally robust learning frameworks to minimize the worst-case expected loss over a prescribed ambiguity set of training distributions quantified via Wasserstein distance. Later, we build on this robust framework to design robust semi-supervised learning over graph methods. The second part of this thesis aspires to fully unleash the potential of next-generation wired and wireless networks, where we design "smart" network entities using (deep) reinforcement learning approaches. Finally, this thesis enhances the power system operation and control. Our contribution is on sustainable distribution grids with high penetration of renewable sources and demand response programs. To account for unanticipated and rapidly changing renewable generation and load consumption scenarios, we specifically delegate reactive power compensation to both utility-owned control devices (e.g., capacitor banks), as well as smart inverters of distributed generation units with cyber-capabilities.Alireza Sadeghiwork_ltekzvvpuncclag5vmbj2qrhauTue, 08 Feb 2022 00:00:00 GMTNovel Concepts and Designs for Adversarial Attacks and Defenses
https://scholar.archive.org/work/qkq76qkqjbhancmo2gxcxghtzi
Albeit displaying remarkable performance across a range of tasks, Deep Neural Networks (DNNs) are highly vulnerable to adversarial examples which are carefully created to deceive these networks. This thesis first demonstrates that DNNs are vulnerable against adversarial attacks even when the attacker is unaware of the model architecture or the training data used to train the model and then proposes a number of novel approaches to improve the robustness of DNNs against challenging adversarial perturbations. Specifically for adversarial attacks, our work highlights how targeted and untargeted adversarial functions can be learned without access to the original data distribution, training mechanism, or label space of an attacked computer vision system. We demonstrate state-of-the-art cross-domain transferability of adversarial perturbations learned from paintings, cartoons, and medical scans to models trained on natural image datasets (such as ImageNet). In this manner, our work highlights an important vulnerability of deep neural networks which makes their deployment challenging in a real-world scenario. Against the threat of these adversarial attacks, we develop novel defense mechanisms that can be deployed with or without retraining the deep neural networks. To this end, we design two plug-and-play defense methods that can protect off-the-shelf pre-trained models without retraining. Specifically, we propose Neural Representation Purifier (NRP) and Local Gradient Smoothing (LGS) to defend against constrained and unconstrained attacks, respectively. NRP learns to purify adversarial noise spread across entire the input image, however, it still struggles against unconstrained attacks where an attacker hides an adversarial sticker preferably in the background without disturbing the original salient image information. We develop a mechanism to smooth local gradients in an input image to stabilize abnormal adversarial patterns introduced by the unconstrained attacks such as an adversarial patch. Robustifying model's para [...]Muzammal Naseer, University, The Australian Nationalwork_qkq76qkqjbhancmo2gxcxghtziSat, 01 Jan 2022 00:00:00 GMTStatistical approach to tagging stellar birth groups in the Milky Way
https://scholar.archive.org/work/4wwvnk3po5hubaoecsxwyztpim
Statistical approach to tagging stellar birth groups in the Milky Way Bridget L. Ratcliffe A major goal of the field of Galactic archeology is to understand the formation and evolution of the Milky Way disk. Stars migrate to different Galactic radii throughout their lifetimes, often leaving little dynamical signature of their initial orbits. Therefore, we need to look at the archaeological record preserved in stellar chemical compositions, which is indicative of their birth environment. In this thesis, we use the measurable properties of stars (chemical compositions and ages) to reconstruct the Milky Way disk's past. First, using hydrodynamical simulations, we find that a star's birth radius and age are linked to its chemical abundances. Subsequently, we learn that even with current-day measurement uncertainty and sample sizes, chemical abundances of Milky Way stars provide a route to reconstructing its formation over time. Extending the insights from hydrodynamical simulations to 30,000 stars observed across the Milky Way disk in the APOGEE survey reveals the importance of using the high-dimensional chemical abundance space. Specifically, we determine that we can use groups of chemically similar stars with 19 measured abundances to trace different underlying formation conditions. Using the high-dimensional abundance data for 10 4 stars from two spectroscopic surveys, APOGEE and GALAH, we empirically describe the chemical abundance trends across a vast radial extent of the Milky Way disk. To do this, we employ a novel approach of quantifying radial variations for individual abundances conditioned on supernovae enrichment history.Bridget Lynn Ratcliffework_4wwvnk3po5hubaoecsxwyztpimD6.2 - Preliminary conclusions about Federated Learning applied to clinical data
https://scholar.archive.org/work/2ibqahhtr5b5ti5wmhzyc5a2em
This report comprises the first contributions from different partners on Federated Learning (FL). After a preliminary introductory section where the fundamental procedures and limitations are described, we detail the well-known mathematical foundation of Federated Learning for convex problems. In this case, we present a key algorithm, Alternating Direction Multipliers Method (ADMM), which is able to implement in a distributed way some fundamental problems such as regression (Ridge and LASSO) and classification (Logistic Regression and Support Vector Machines (SVM)). This procedure shares the fundamental approach of FL, which consists of performing some local processing, sharing some intermediate information and updating the local information with some global innovation. In a second step we introduce the extension of this approach to non-convex problems using Bayesian Neural Networks (BNN) where the update is based on the cooperative construction of the posterior of weights from different architectures. Several sections follow where different partners provide different contributions describing our first initiatives on the topic. Some preliminary code from all partners has been uploaded to a common repository to start creating a pool of methods and tools to foster incoming synergies.Federico Álvarez, Santiago Zazo, Juan Parras, Alejandro Almodóvar, Patricia Alonso, Enrico Giampieri, Gastone Castellani, Lorenzo Sani, Cesare Rollo, Tiziana Sanavia, Anders Krogh, Íñigo Prada-Luengo, Alexandros Kanterakis, Stelios Sfakianakis, Francesco Cremonesiwork_2ibqahhtr5b5ti5wmhzyc5a2emFri, 31 Dec 2021 00:00:00 GMTImage Segmentation of Bacterial Cells in Biofilms
https://scholar.archive.org/work/muca6lzgibhtdgm4awyvqiigiu
Bakterielle Biofilme sind drei-dimensionale Zellcluster, welche ihre eigene Matrix produzieren. Die selbst-produzierte Matrix bietet den Zellen einen gemeinschaftlichen Schutz vor äußeren Stressfaktoren. Diese Stressfaktoren können abiotischer Natur sein wie z.B. Temperatur- und Nährstoff\- schwankungen, oder aber auch biotische Faktoren wie z.B. Antibiotikabehandlung oder Bakteriophageninfektionen. Dies führt dazu, dass einzelne Zelle innerhalb der mikrobiologischen Gemeinschaften eine erhöhte Widerstandsfähigkeit aufweisen und eine große Herausforderung für Medizin und technische Anwendungen darstellen. Um Biofilme wirksam zu bekämpfen, muss man die dem Wachstum und Entwicklung zugrundeliegenden Mechanismen entschlüsseln. Aufgrund der hohen Zelldichte innerhalb der Gemeinschaften sind die Mechanismen nicht räumlich und zeitlich invariant, sondern hängen z.B. von Metabolit-, Nährstoff- und Sauerstoffgradienten ab. Daher ist es für die Beschreibung unabdingbar Beobachtungen auf Einzelzellebene durchzuführen. Für die nicht-invasive Untersuchung von einzelnen Zellen innerhalb eines Biofilms ist man auf konfokale Fluoreszenzmikroskopie angewiesen. Um aus den gesammelten, drei-dimensionalen Bilddaten Zelleigenschaften zu extrahieren, ist die Erkennung von den jeweiligen Zellen erforderlich. Besonders die digitale Rekonstruktion der Zellmorphologie spielt dabei eine große Rolle. Diese erhält man über die Segmentierung der Bilddaten. Dabei werden einzelne Bildelemente den abgebildeten Objekten zugeordnet. Damit lassen sich die einzelnen Objekte voneinander unterscheiden und deren Eigenschaften extrahieren. Im ersten Teil dieser Arbeit wird ein benutzerfreundliches Computerprogramm vorgestellt, welches die Segmentierung und Analyse von Fluoreszenzmikroskopiedaten wesentlich vereinfacht. Es stellt eine umfangreiche Auswahl an traditionellen Segmentieralgorithmen, Parameterberechnungen und Visualisierungsmöglichkeiten zur Verfügung. Alle Funktionen sind ohne Programmierkenntnisse zugänglich, sodass sie einer großen Gruppe [...]Eric Jelli, Physik, Drescher, Knut (Prof. Dr.)work_muca6lzgibhtdgm4awyvqiigiuThu, 02 Dec 2021 00:00:00 GMTAn Introduction to Probabilistic Programming
https://scholar.archive.org/work/dadto6xdezehzodoaioe6qfacq
This book is a graduate-level introduction to probabilistic programming. It not only provides a thorough background for anyone wishing to use a probabilistic programming system, but also introduces the techniques needed to design and build these systems. It is aimed at people who have an undergraduate-level understanding of either or, ideally, both probabilistic machine learning and programming languages. We start with a discussion of model-based reasoning and explain why conditioning is a foundational computation central to the fields of probabilistic machine learning and artificial intelligence. We then introduce a first-order probabilistic programming language (PPL) whose programs correspond to graphical models with a known, finite, set of random variables. In the context of this PPL we introduce fundamental inference algorithms and describe how they can be implemented. We then turn to higher-order probabilistic programming languages. Programs in such languages can define models with dynamic computation graphs, which may not instantiate the same set of random variables in each execution. Inference requires methods that generate samples by repeatedly evaluating the program. Foundational algorithms for this kind of language are discussed in the context of an interface between program executions and an inference controller. Finally we consider the intersection of probabilistic and differentiable programming. We begin with a discussion of automatic differentiation, and how it can be used to implement efficient inference methods based on Hamiltonian Monte Carlo. We then discuss gradient-based maximum likelihood estimation in programs that are parameterized using neural networks, how to amortize inference using by learning neural approximations to the program posterior, and how language features impact the design of deep probabilistic programming systems.Jan-Willem van de Meent, Brooks Paige, Hongseok Yang, Frank Woodwork_dadto6xdezehzodoaioe6qfacqTue, 19 Oct 2021 00:00:00 GMTUncertainty-aware Salient Object Detection
https://scholar.archive.org/work/qpse66alsvdqnf7briebnmir3a
Saliency detection models are trained to discover the region(s) of an image that attract human attention. According to whether depth data is used, static image saliency detection models can be divided into RGB image saliency detection models, and RGB-D image saliency detection models. The former predict salient regions of the RGB image, while the latter take both the RGB image and the depth data as input. Conventional saliency prediction models typically learn a deterministic mapping from images to the corresponding ground truth saliency maps without modeling the uncertainty of predictions, following the supervised learning pipeline. This thesis is dedicated to learning a conditional distribution over saliency maps, given an input image, and modeling the uncertainty of predictions. For RGB-D saliency detection, we present the first generative model based framework to achieve uncertainty-aware prediction. Our framework includes two main models: 1) a generator model and 2) an inference model. The generator model is an encoder-decoder saliency network. To infer the latent variable, we introduce two different solutions: i) a Conditional Variational Auto-encoder with an extra encoder to approximate the posterior distribution of the latent variable; and ii) an Alternating Back-Propagation technique, which directly samples the latent variable from the true posterior distribution. One drawback of above model is that it fails to explicitly model the connection between RGB image and depth data to achieve effective cooperative learning. We further introduce a novel latent variable model based complementary learning framework to explicitly model the complementary information between the two modes, namely the RGB mode and depth mode. Specifically, we first design a regularizer using mutual-information minimization to reduce the redundancy between appearance features from RGB and geometric features from depth in the latent space. Then we fuse the latent features of each mode to achieve multi-modal feature fusion. Extensive exp [...]Jing Zhang, University, The Australian Nationalwork_qpse66alsvdqnf7briebnmir3aSun, 03 Oct 2021 00:00:00 GMTAdvances in adversarial attacks and defenses in computer vision: A survey
https://scholar.archive.org/work/6n7qe7hhurfrphkshhqnr2ug2i
Deep Learning (DL) is the most widely used tool in the contemporary field of computer vision. Its ability to accurately solve complex problems is employed in vision research to learn deep neural models for a variety of tasks, including security critical applications. However, it is now known that DL is vulnerable to adversarial attacks that can manipulate its predictions by introducing visually imperceptible perturbations in images and videos. Since the discovery of this phenomenon in 2013~[1], it has attracted significant attention of researchers from multiple sub-fields of machine intelligence. 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. Many of those contributions have inspired new directions in this area, which has matured significantly since witnessing the first generation methods. Hence, as a legacy sequel of [2], this literature review focuses on the advances in this area since 2018. To ensure authenticity, we mainly consider peer-reviewed contributions published in the prestigious sources of computer vision and machine learning research. Besides a comprehensive literature review, the article also provides concise definitions of technical terminologies for non-experts in this domain. Finally, this article discusses challenges and future outlook of this direction based on the literature reviewed herein and [2].Naveed Akhtar, Ajmal Mian, Navid Kardan, Mubarak Shahwork_6n7qe7hhurfrphkshhqnr2ug2iThu, 02 Sep 2021 00:00:00 GMTIdentifying and Exploiting Structures for Reliable Deep Learning
https://scholar.archive.org/work/aiftbjpfmfg3fghj7n4mefjf5u
Deep learning research has recently witnessed an impressively fast-paced progress in a wide range of tasks including computer vision, natural language processing, and reinforcement learning. The extraordinary performance of these systems often gives the impression that they can be used to revolutionise our lives for the better. However, as recent works point out, these systems suffer from several issues that make them unreliable for use in the real world, including vulnerability to adversarial attacks (Szegedy et al. [248]), tendency to memorise noise (Zhang et al. [292]), being over-confident on incorrect predictions (miscalibration) (Guo et al. [99]), and unsuitability for handling private data (Gilad-Bachrach et al. [88]). In this thesis, we look at each of these issues in detail, investigate their causes, and propose computationally cheap algorithms for mitigating them in practice. To do this, we identify structures in deep neural networks that can be exploited to mitigate the above causes of unreliability of deep learning algorithms.Amartya Sanyalwork_aiftbjpfmfg3fghj7n4mefjf5uMon, 16 Aug 2021 00:00:00 GMT