We investigate replicable learning algorithms. Informally a learning algorithm is replicable if the algorithm outputs the same canonical hypothesis over multiple runs with high probability, even when different runs ob...
详细信息
ISBN:
(纸本)9781713899921
We investigate replicable learning algorithms. Informally a learning algorithm is replicable if the algorithm outputs the same canonical hypothesis over multiple runs with high probability, even when different runs observe a different set of samples from the unknown data distribution. In general, such a strong notion of replicability is not achievable. Thus we consider two feasible notions of replicability called list replicability and certificate replicability. Intuitively, these notions capture the degree of (non) replicability. The goal is to design learning algorithms with optimal list and certificate complexities while minimizing the sample complexity. Our contributions are the following. - We first study the learning task of estimating the biases of d coins, up to an additive error of epsilon, by observing samples. For this task, we design a (d + 1)-list replicable algorithm. To complement this result, we establish that the list complexity is optimal, i.e there are no learning algorithms with a list size smaller than d + 1 for this task. We also design learning algorithms with certificate complexity (O) over tilde (log d). The sample complexity of both these algorithms is (O) over tilde (d(2)/epsilon(2)) where e is the approximation error parameter (for a constant error probability). - In the PAC model, we show that any hypothesis class that is learnable with d-nonadaptive statistical queries can be learned via a (d + 1)-list replicable algorithm and also via a (O) over tilde (log d)-certificate replicable algorithm. The sample complexity of both these algorithms is (O) over tilde (d(2)/nu(2)) where. is the approximation error of the statistical query. We also show that for the concept class d-THRESHOLD, the list complexity is exactly d + 1 with respect to the uniform distribution. To establish our upper bound results we use rounding schemes induced by geometric partitions with certain properties. We use Sperner/KKM Lemma to establish the lower bound results.
This article explores the application of deep learning (DL) algorithms in power system load forecasting. With the continuous advancement of the construction of new power systems, traditional load forecasting models de...
详细信息
Traffic sign recognition is crucial for the safe and efficient operation of autonomous vehicles. While previous research has primarily focused on traffic sign recognition in foreign countries, these studies often face...
详细信息
ISBN:
(纸本)9798350352368
Traffic sign recognition is crucial for the safe and efficient operation of autonomous vehicles. While previous research has primarily focused on traffic sign recognition in foreign countries, these studies often face limitations such as differing traffic sign designs, language barriers in textual information, and varying environmental conditions. In this paper, we propose a traffic sign detection and recognition system tailored for Malaysia, utilizing Convolutional Neural Networks (CNNs) and optical Character Recognition (OCR). In this paper, we propose a traffic sign detection and recognition system utilizing You Only Look Once (YOLO) v8 for object detection and EasyOCR to process textual information on selected traffic signs. Our system achieves a mean Average Precision (mAP) of 0.824 and an average processing time of 1.2 seconds per frame, which is comparable to existing literature. Furthermore, the complexity of our method is significantly reduced, enhancing its potential for real-time processing applications, as evidenced by its efficient processing time.
The interference cancellation problem for the satellite navigation systems is considered. The space-frequency adaptive processingalgorithms that use the digital beamforming technique are described. Different beamform...
详细信息
In an internet of vehicle (Iov) scenario, vehicular edge computing (vEC) exploits the computing capabilities of the vehicles and roadside unit (RSU) to enhance the task processing capabilities of the vehicles. Resourc...
详细信息
In an internet of vehicle (Iov) scenario, vehicular edge computing (vEC) exploits the computing capabilities of the vehicles and roadside unit (RSU) to enhance the task processing capabilities of the vehicles. Resource management is essential to the performance improvement of the vEC system. In this paper, we propose a joint task offloading and resource allocation scheme to minimize the total task processing delay of all the vehicles through task scheduling, channel allocation, and computing resource allocation for the vehicles and RSU. Different from the existing works, our scheme: 1) considers task diversity by profiling the tasks of the vehicles by multiple attributes including data size, computation amount, delay tolerance, and task type;2) considers vehicle classification by dividing the vehicles into 4 sets according to whether they have task offloading requirements or provide task processing services;3) considers task processing flexibility by deciding for each vehicle to process its tasks locally, to offload the tasks to the RSU via v2I (vehicle to Infrastructure) connections, or to the other vehicles via v2v (vehicle to vehicle) connections. An algorithm based on the Generalized Benders Decomposition (GBD) and Reformulation Linearization (RL) methods is designed to optimally solve the optimization problem. A heuristic algorithm is also designed to provide the sub-optimal solution with low computational complexity. We analyze the convergence and complexity of the proposed algorithms and conduct extensive simulations in 6 scenarios. The simulation results demonstrate the superiority of our scheme in comparison with 4 other schemes.
Accurate positioning in shallow water regions is crucial for various underwater applications. TDOA algorithms estimate the signal source location by measuring the time differences of signal arrival at multiple receive...
详细信息
The research proposes a system for recording and processing the spatial characteristics of the laser beam to detect the thermal convective flow appearing already at an early stage of the fire. The work is carried out ...
详细信息
ISBN:
(纸本)9783031214370;9783031214387
The research proposes a system for recording and processing the spatial characteristics of the laser beam to detect the thermal convective flow appearing already at an early stage of the fire. The work is carried out within the framework of the development of new effective methods and technical means of early fire detection. Methods of computer processing of spatial characteristics of a laser beam, allowing to detection of the appearance of a thermal convective flow, are investigated. The efficiency of the correlation method of processing spatial characteristics has been investigated. A processing method, reflecting a pixel-by-pixel difference of counts of the intensity distribution of the beam profile, is offered. Their efficiency has been compared.
Addressing the growing challenges posed by increasing vehicles and stringent traffic violations, many vehicle number plates go undetected while passing through toll booths and traffic lights. Thus it creates the need ...
详细信息
This paper proposes algorithmic and technical solutions aimed at improving the accuracy of UAv navigation parameters without altering the onboard navigation system's structure. The main focus is on introducing add...
详细信息
Microbial OD600 measurement is a crucial experimental technique in microbiological and molecular biological research, evaluating microbial growth by measuring optical density at 600 nm wavelength. While convenient, th...
详细信息
暂无评论