Abstract
Conventional frequentist FL schemes are known to yield overconfident decisions. Bayesian FL addresses this issue by allowing agents to process and exchange uncertainty information encoded in distributions over the model parameters. However, this comes at the cost of a larger per-iteration communication overhead. This letter investigates whether Bayesian FL can still provide advantages in terms of calibration when constraining communication bandwidth. We present compressed particle-based Bayesian FL protocols for FL and federated "unlearning" that apply quantization and sparsification across multiple particles. The experimental results confirm that the benefits of Bayesian FL are robust to bandwidth constraints.
Keyword: overconfidence
There is no result
Keyword: confidence
PROB-SLAM: Real-time Visual SLAM Based on Probabilistic Graph Optimization
Authors: Xianwei Meng, Bonian Li
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Traditional SLAM algorithms are typically based on artificial features, which lack high-level information. By introducing semantic information, SLAM can own higher stability and robustness rather than purely hand-crafted features. However, the high uncertainty of semantic detection networks prohibits the practical functionality of high-level information. To solve the uncertainty property introduced by semantics, this paper proposed a novel probability map based on the Gaussian distribution assumption. This map transforms the semantic binary object detection into probability results, which help establish a probabilistic data association between artificial features and semantic info. Through our algorithm, the higher confidence will be given higher weights in each update step while the edge of the detection area will be endowed with lower confidence. Then the uncertainty is undermined and has less effect on nonlinear optimization. The experiments are carried out in the TUM RGBD dataset, results show that our system improves ORB-SLAM2 by about 15% in indoor environments' errors. We have demonstrated that the method can be successfully applied to environments containing dynamic objects.
Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models
Authors: Manli Shu, Weili Nie, De-An Huang, Zhiding Yu, Tom Goldstein, Anima Anandkumar, Chaowei Xiao
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
Pre-trained vision-language models (e.g., CLIP) have shown promising zero-shot generalization in many downstream tasks with properly designed text prompts. Instead of relying on hand-engineered prompts, recent works learn prompts using the training data from downstream tasks. While effective, training on domain-specific data reduces a model's generalization capability to unseen new domains. In this work, we propose test-time prompt tuning (TPT), a method that can learn adaptive prompts on the fly with a single test sample. For image classification, TPT optimizes the prompt by minimizing the entropy with confidence selection so that the model has consistent predictions across different augmented views of each test sample. In evaluating generalization to natural distribution shifts, TPT improves the zero-shot top-1 accuracy of CLIP by 3.6% on average, surpassing previous prompt tuning approaches that require additional task-specific training data. In evaluating cross-dataset generalization with unseen categories, TPT performs on par with the state-of-the-art approaches that use additional training data. Project page: https://azshue.github.io/TPT.
Keyword: scaling
SQL and NoSQL Databases Software architectures performance analysis and assessments -- A Systematic Literature review
Authors: Wisal Khan, Teerath Kumar, Zhang Cheng, Kislay Raj, Arunabha M Roy, Bin Luo
Subjects: Databases (cs.DB); Artificial Intelligence (cs.AI); Programming Languages (cs.PL)
Abstract
Context: The efficient processing of Big Data is a challenging task for SQL and NoSQL Databases, where competent software architecture plays a vital role. The SQL Databases are designed for structuring data and supporting vertical scalability. In contrast, horizontal scalability is backed by NoSQL Databases and can process sizeable unstructured Data efficiently. One can choose the right paradigm according to the organisation's needs; however, making the correct choice can often be challenging. The SQL and NoSQL Databases follow different architectures. Also, the mixed model is followed by each category of NoSQL Databases. Hence, data movement becomes difficult for cloud consumers across multiple cloud service providers (CSPs). In addition, each cloud platform IaaS, PaaS, SaaS, and DBaaS also monitors various paradigms. Objective: This systematic literature review (SLR) aims to study the related articles associated with SQL and NoSQL Database software architectures and tackle data portability and Interoperability among various cloud platforms. State of the art presented many performance comparison studies of SQL and NoSQL Databases by observing scaling, performance, availability, consistency and sharding characteristics. According to the research studies, NoSQL Database designed structures can be the right choice for big data analytics, while SQL Databases are suitable for OLTP Databases. The researcher proposes numerous approaches associated with data movement in the cloud. Platform-based APIs are developed, which makes users' data movement difficult. Therefore, data portability and Interoperability issues are noticed during data movement across multiple CSPs. To minimize developer efforts and Interoperability, Unified APIs are demanded to make data movement relatively more accessible among various cloud platforms.
Understanding Deep Neural Function Approximation in Reinforcement Learning via $ε$-Greedy Exploration
Abstract
This paper provides a theoretical study of deep neural function approximation in reinforcement learning (RL) with the $\epsilon$-greedy exploration under the online setting. This problem setting is motivated by the successful deep Q-networks (DQN) framework that falls in this regime. In this work, we provide an initial attempt on theoretical understanding deep RL from the perspective of function class and neural networks architectures (e.g., width and depth) beyond the "linear" regime. To be specific, we focus on the value based algorithm with the $\epsilon$-greedy exploration via deep (and two-layer) neural networks endowed by Besov (and Barron) function spaces, respectively, which aims at approximating an $\alpha$-smooth Q-function in a $d$-dimensional feature space. We prove that, with $T$ episodes, scaling the width $m = \widetilde{\mathcal{O}}(T^{\frac{d}{2\alpha + d}})$ and the depth $L=\mathcal{O}(\log T)$ of the neural network for deep RL is sufficient for learning with sublinear regret in Besov spaces. Moreover, for a two layer neural network endowed by the Barron space, scaling the width $\Omega(\sqrt{T})$ is sufficient. To achieve this, the key issue in our analysis is how to estimate the temporal difference error under deep neural function approximation as the $\epsilon$-greedy exploration is not enough to ensure "optimism". Our analysis reformulates the temporal difference error in an $L^2(\mathrm{d}\mu)$-integrable space over a certain averaged measure $\mu$, and transforms it to a generalization problem under the non-iid setting. This might have its own interest in RL theory for better understanding $\epsilon$-greedy exploration in deep RL.
Arithmetic circuit tensor networks, multivariable function representation, and high-dimensional integration
Abstract
Many computational problems can be formulated in terms of high-dimensional functions. Simple representations of such functions and resulting computations with them typically suffer from the "curse of dimensionality", an exponential cost dependence on dimension. Tensor networks provide a way to represent certain classes of high-dimensional functions with polynomial memory. This results in computations where the exponential cost is ameliorated or in some cases, removed, if the tensor network representation can be obtained. Here, we introduce a direct mapping from the arithmetic circuit of a function to arithmetic circuit tensor networks, avoiding the need to perform any optimization or functional fit. We demonstrate the power of the circuit construction in examples of multivariable integration on the unit hypercube in up to 50 dimensions, where the complexity of integration can be understood from the circuit structure. We find very favorable cost scaling compared to quasi-Monte-Carlo integration for these cases, and further give an example where efficient quasi-Monte-Carlo cannot be theoretically performed without knowledge of the underlying tensor network circuit structure.
Keyword: calibration
Uncertainty-Aware Visual Perception for Safe Motion Planning
Authors: Ralf Römer, Armin Lederer, Samuel Tesfazgi, Sandra Hirche
Subjects: Robotics (cs.RO); Systems and Control (eess.SY)
Abstract
For safe operation, a robot must be able to avoid collisions in uncertain environments. Existing approaches for motion planning with uncertainties often make conservative assumptions about Gaussianity and the obstacle geometry. While visual perception can deliver a more accurate representation of the environment, its use for safe motion planning is limited by the inherent miscalibration of neural networks and the challenge of obtaining adequate datasets. In order to address these imitations, we propose to employ ensembles of deep semantic segmentation networks trained with systematically augmented datasets to ensure reliable probabilistic occupancy information. For avoiding conservatism during motion planning, we directly employ the probabilistic perception via a scenario-based path planning approach. A velocity scheduling scheme is applied to the path to ensure a safe motion despite tracking inaccuracies. We demonstrate the effectiveness of the systematic data augmentation in combination with deep ensembles and the proposed scenario-based planning approach in comparisons to state-of-the-art methods and validate our framework in an experiment involving a human hand.
Compressed Particle-Based Federated Bayesian Learning and Unlearning
Authors: Jinu Gong, Osvaldo Simeone, Joonhyuk Kang
Subjects: Machine Learning (cs.LG); Signal Processing (eess.SP)
Abstract
Conventional frequentist FL schemes are known to yield overconfident decisions. Bayesian FL addresses this issue by allowing agents to process and exchange uncertainty information encoded in distributions over the model parameters. However, this comes at the cost of a larger per-iteration communication overhead. This letter investigates whether Bayesian FL can still provide advantages in terms of calibration when constraining communication bandwidth. We present compressed particle-based Bayesian FL protocols for FL and federated "unlearning" that apply quantization and sparsification across multiple particles. The experimental results confirm that the benefits of Bayesian FL are robust to bandwidth constraints.
Online Marker-free Extrinsic Camera Calibration using Person Keypoint Detections
Authors: Bastian Pätzold, Simon Bultmann, Sven Behnke
Subjects: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
Abstract
Calibration of multi-camera systems, i.e. determining the relative poses between the cameras, is a prerequisite for many tasks in computer vision and robotics. Camera calibration is typically achieved using offline methods that use checkerboard calibration targets. These methods, however, often are cumbersome and lengthy, considering that a new calibration is required each time any camera pose changes. In this work, we propose a novel, marker-free online method for the extrinsic calibration of multiple smart edge sensors, relying solely on 2D human keypoint detections that are computed locally on the sensor boards from RGB camera images. Our method assumes the intrinsic camera parameters to be known and requires priming with a rough initial estimate of the camera poses. The person keypoint detections from multiple views are received at a central backend where they are synchronized, filtered, and assigned to person hypotheses. We use these person hypotheses to repeatedly solve optimization problems in the form of factor graphs. Given suitable observations of one or multiple persons traversing the scene, the estimated camera poses converge towards a coherent extrinsic calibration within a few minutes. We evaluate our approach in real-world settings and show that the calibration with our method achieves lower reprojection errors compared to a reference calibration generated by an offline method using a traditional calibration target.
Omnipredictors for Constrained Optimization
Authors: Lunjia Hu, Inbal Livni-Navon, Omer Reingold, Chutong Yang
Subjects: Machine Learning (cs.LG); Computers and Society (cs.CY); Data Structures and Algorithms (cs.DS); Optimization and Control (math.OC)
Abstract
The notion of omnipredictors (Gopalan, Kalai, Reingold, Sharan and Wieder ITCS 2021), suggested a new paradigm for loss minimization. Rather than learning a predictor based on a known loss function, omnipredictors can easily be post-processed to minimize any one of a rich family of loss functions compared with the loss of a class $C$. It has been shown that such omnipredictors exist and are implied (for all convex and Lipschitz loss functions) by the notion of multicalibration from the algorithmic fairness literature. Nevertheless, it is often the case that the action selected must obey some additional constraints (such as capacity or parity constraints). In itself, the original notion of omnipredictors does not apply in this well-motivated and heavily studied the context of constrained loss minimization. In this paper, we introduce omnipredictors for constrained optimization and study their complexity and implications. The notion that we introduce allows the learner to be unaware of the loss function that will be later assigned as well as the constraints that will be later imposed, as long as the subpopulations that are used to define these constraints are known. The paper shows how to obtain omnipredictors for constrained optimization problems, relying on appropriate variants of multicalibration. For some interesting constraints and general loss functions and for general constraints and some interesting loss functions, we show how omnipredictors are implied by a variant of multicalibration that is similar in complexity to standard multicalibration. We demonstrate that in the general case, standard multicalibration is insufficient and show that omnipredictors are implied by multicalibration with respect to a class containing all the level sets of hypotheses in $C$. We also investigate the implications when the constraints are group fairness notions.
Keyword: out of distribution detection
There is no result
Keyword: out-of-distribution detection
There is no result
Keyword: expected calibration error
There is no result
Keyword: overconfident
Compressed Particle-Based Federated Bayesian Learning and Unlearning
Keyword: overconfidence
There is no result
Keyword: confidence
PROB-SLAM: Real-time Visual SLAM Based on Probabilistic Graph Optimization
Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models
Keyword: scaling
SQL and NoSQL Databases Software architectures performance analysis and assessments -- A Systematic Literature review
Understanding Deep Neural Function Approximation in Reinforcement Learning via $ε$-Greedy Exploration
Arithmetic circuit tensor networks, multivariable function representation, and high-dimensional integration
Keyword: calibration
Uncertainty-Aware Visual Perception for Safe Motion Planning
Compressed Particle-Based Federated Bayesian Learning and Unlearning
Online Marker-free Extrinsic Camera Calibration using Person Keypoint Detections
Omnipredictors for Constrained Optimization