scholarly journals DACBench: A Benchmark Library for Dynamic Algorithm Configuration

2021 ◽  
Author(s):  
Theresa Eimer ◽  
André Biedenkapp ◽  
Maximilian Reimer ◽  
Steven Adriansen ◽  
Frank Hutter ◽  
...  
Keyword(s):  
Deep Learning ◽  
Evolutionary Computation ◽  
Automatic Documentation ◽  
Ai Planning ◽  
Dynamic Algorithm ◽  
Empirical Results ◽  
Broad Applicability ◽  
New Methods ◽  
Algorithm Configuration

Dynamic Algorithm Configuration (DAC) aims to dynamically control a target algorithm's hyperparameters in order to improve its performance. Several theoretical and empirical results have demonstrated the benefits of dynamically controlling hyperparameters in domains like evolutionary computation, AI Planning or deep learning. Replicating these results, as well as studying new methods for DAC, however, is difficult since existing benchmarks are often specialized and incompatible with the same interfaces. To facilitate benchmarking and thus research on DAC, we propose DACBench, a benchmark library that seeks to collect and standardize existing DAC benchmarks from different AI domains, as well as provide a template for new ones. For the design of DACBench, we focused on important desiderata, such as (i) flexibility, (ii) reproducibility, (iii) extensibility and (iv) automatic documentation and visualization. To show the potential, broad applicability and challenges of DAC, we explore how a set of six initial benchmarks compare in several dimensions of difficulty.

scholarly journals KeyNet: An Asymmetric Key-Style Framework for Watermarking Deep Learning Models

2021 ◽  
Vol 11 (3) ◽  
pp. 999
Author(s):  
Najeeb Moharram Jebreel ◽  
Josep Domingo-Ferrer ◽  
David Sánchez ◽  
Alberto Blanco-Justicia
Keyword(s):  
Deep Learning ◽  
Intellectual Property ◽  
High Fidelity ◽  
Data Sets ◽  
Learning Models ◽  
Robust Watermarking ◽  
Empirical Results ◽  
Private Key ◽  
Extensive Evaluation ◽  
Original Classification

Many organizations devote significant resources to building high-fidelity deep learning (DL) models. Therefore, they have a great interest in making sure the models they have trained are not appropriated by others. Embedding watermarks (WMs) in DL models is a useful means to protect the intellectual property (IP) of their owners. In this paper, we propose KeyNet, a novel watermarking framework that satisfies the main requirements for an effective and robust watermarking. In KeyNet, any sample in a WM carrier set can take more than one label based on where the owner signs it. The signature is the hashed value of the owner’s information and her model. We leverage multi-task learning (MTL) to learn the original classification task and the watermarking task together. Another model (called the private model) is added to the original one, so that it acts as a private key. The two models are trained together to embed the WM while preserving the accuracy of the original task. To extract a WM from a marked model, we pass the predictions of the marked model on a signed sample to the private model. Then, the private model can provide the position of the signature. We perform an extensive evaluation of KeyNet’s performance on the CIFAR10 and FMNIST5 data sets and prove its effectiveness and robustness. Empirical results show that KeyNet preserves the utility of the original task and embeds a robust WM.

Reliability Analysis Using Deep Learning

2018 ◽  
Author(s):  
Chong Chen ◽  
Ying Liu ◽  
Xianfang Sun ◽  
Shixuan Wang ◽  
Carla Di Cairano-Gilfedder ◽  
...  
Keyword(s):  
Data Mining ◽  
Deep Learning ◽  
Reliability Analysis ◽  
General Principle ◽  
Preventive Maintenance ◽  
Maintenance Cost ◽  
Empirical Results ◽  
Time Between Failures ◽  
Learning Techniques ◽  
Deep Learning Model

Over the last few decades, reliability analysis has gained more and more attention as it can be beneficial in lowering the maintenance cost. Time between failures (TBF) is an essential topic in reliability analysis. If the TBF can be accurately predicted, preventive maintenance can be scheduled in advance in order to avoid critical failures. The purpose of this paper is to research the TBF using deep learning techniques. Deep learning, as a tool capable of capturing the highly complex and nonlinearly patterns, can be a useful tool for TBF prediction. The general principle of how to design deep learning model was introduced. By using a sizeable amount of automobile TBF dataset, we conduct an experiential study on TBF prediction by deep learning and several data mining approaches. The empirical results show the merits of deep learning in performance but comes with cost of high computational load.

scholarly journals Pre-trained deep learning models in automatic COVID-19 diagnosis

2021 ◽  
Vol 22 (3) ◽  
pp. 1540
Author(s):  
Ahmed Wasif Reza ◽  
Md Mahamudul Hasan ◽  
Nazla Nowrin ◽  
Mir Moynuddin Ahmed Shibly
Keyword(s):  
Deep Learning ◽  
Open Source ◽  
Healthcare Professionals ◽  
Automated System ◽  
Primary Concern ◽  
X Rays ◽  
Learning Models ◽  
Efficient System ◽  
Empirical Results ◽  
History Of

Coronavirus Disease (COVID-19) is a devastating pandemic in the history of mankind. It is a highly contagious flu that can spread from human to human without revealing any symptoms. For being so contagious, detecting patients with it and isolating them has become the primary concern for healthcare professionals. This study presented an alternative way to identify COVID-19 patients by doing an automatic examination of chest X-rays of the patients. To develop such an efficient system, six pre-trained deep learning models were used. Those models were: VGG16, InceptionV3, Xception, DenseNet201, InceptionResNetV2, and EfficientNetB4. Those models were developed on two open-source datasets that have chest X-rays of patients diagnosed with COVID-19. Among the models, EfficientNetB4 achieved better performances on both datasets with 96% and 97% of accuracies. The empirical results were also exemplary. This type of automated system can help us fight this dangerous virus outbreak.

scholarly journals An Ensemble Deep Dynamic Algorithm (EDDA) to Predict the Heart Disease

2021 ◽  
pp. 105-111
Author(s):  
J. Nageswara Rao ◽  
R. Satya Prasad
Keyword(s):  
Heart Disease ◽  
Deep Learning ◽  
Disease Prediction ◽  
Dynamic Algorithm ◽  
Risk Of Death ◽  
Deep Boltzmann Machine ◽  
Comparative Results ◽  
Sensitivity Specificity ◽  
Accuracy Of Prediction ◽  
Automatic Systems

Nowadays heart disease becomes more complicated to every human being. Machine Learning and Deep Learning plays the major role in processing the automatic systems. Prediction of heart disease is most difficult task because many algorithms perform limited operations. The aim of the paper is to increase the accuracy and prediction values. Various heart disease datasets are available for the research. Deep Learning (DL) algorithms play the major role in prediction of heart disease. Prediction can be done in the early stages to reduce the risk of death for the humans. In this paper, An Ensemble Deep Dynamic Algorithm (EDDA) is introduced to increase the accuracy of prediction values. The EDDA follows the some steps to process the prediction of heart disease. The steps are as follows: Linear Regression and Deep Boltzmann Machine (DBM) is applied on the selected dataset. Performance is calculated in terms of sensitivity, specificity and accuracy are shown with the comparative results.

Integrating Evolutionary Computation Components in Ant Colony Optimization

2005 ◽  
pp. 148-180 ◽  
Author(s):  
Sergio Alonso ◽  
Oscar Cordon ◽  
Iñaki Fernández de Viana ◽  
Francisco Herrera
Keyword(s):  
Evolutionary Computation ◽  
Ant Colony Optimization ◽  
Ant Colony ◽  
Ant System ◽  
Empirical Results ◽  
New Models ◽  
Similarities And Differences

This chapter introduces two different ways to integrate Evolutionary Computation Components in Ant Colony Optimization (ACO) Meta-heuristic. First of all, the ACO meta-heuristic is introduced and compared to Evolutionary Computation to notice their similarities and differences. Then two new models of ACO algorithms that include some Evolutionary Computation concepts (Best-Worst Ant System and exchange of memoristic information in parallel ACO algorithms) are presented with some empirical results that show improvements in the quality of the solutions when compared with more basic and classical approaches.

scholarly journals Automated Algorithm Selection: Survey and Perspectives

2019 ◽  
Vol 27 (1) ◽  
pp. 3-45 ◽  
Author(s):  
Pascal Kerschke ◽  
Holger H. Hoos ◽  
Frank Neumann ◽  
Heike Trautmann
Keyword(s):  
State Of The Art ◽  
Combinatorial Problems ◽  
The State ◽  
Problem Instance ◽  
Ai Planning ◽  
Algorithm Selection ◽  
Automated Algorithm ◽  
Continuous Problems ◽  
Algorithm Configuration ◽  
The One

It has long been observed that for practically any computational problem that has been intensely studied, different instances are best solved using different algorithms. This is particularly pronounced for computationally hard problems, where in most cases, no single algorithm defines the state of the art; instead, there is a set of algorithms with complementary strengths. This performance complementarity can be exploited in various ways, one of which is based on the idea of selecting, from a set of given algorithms, for each problem instance to be solved the one expected to perform best. The task of automatically selecting an algorithm from a given set is known as the per-instance algorithm selection problem and has been intensely studied over the past 15 years, leading to major improvements in the state of the art in solving a growing number of discrete combinatorial problems, including propositional satisfiability and AI planning. Per-instance algorithm selection also shows much promise for boosting performance in solving continuous and mixed discrete/continuous optimisation problems. This survey provides an overview of research in automated algorithm selection, ranging from early and seminal works to recent and promising application areas. Different from earlier work, it covers applications to discrete and continuous problems, and discusses algorithm selection in context with conceptually related approaches, such as algorithm configuration, scheduling, or portfolio selection. Since informative and cheaply computable problem instance features provide the basis for effective per-instance algorithm selection systems, we also provide an overview of such features for discrete and continuous problems. Finally, we provide perspectives on future work in the area and discuss a number of open research challenges.

scholarly journals Evolutionary Computation for Multifaceted Web Service Composition

2021 ◽  
Author(s):  
◽  
Alexandre Sawczuk da Silva
Keyword(s):  
Local Search ◽  
Evolutionary Computation ◽  
Web Service ◽  
Service Composition ◽  
Web Service Composition ◽  
Ai Planning ◽  
Multi Objective ◽  
Composition Problem ◽  
Local Search Operator

<p>Automated Web service composition is one of the holy grails of service-oriented computing, since it allows users to create an application simply by specifying the inputs the resulting application should require, the outputs it should produce, and any constraints it should observe. The composition problem has been handled using a variety of techniques, from AI planning to optimisation algorithms, however no work so far has focused on handling multiple composition facets simultaneously, producing solutions that: (1) are fully functional (i.e. fully executable, with semantically-matched inputs and outputs), (2) employ a variety of composition constructs (e.g. sequential, parallel, and choice constructs), and (3) are optimised according to non-functional Quality of Service (QoS) measurements. The overall goal of this thesis is to propose hybrid Web service composition approaches that consider elements from all three facets described above when generating solutions. These approaches combine elements of AI planning and of Evolutionary Computation to allow for the creation of compositions that meet all of these requirements.  Firstly, this thesis proposes two novel approaches for Web service composition with direct representations. The first one is a tree-based approach where the leaf nodes are the atomic services included in the composition and the inner nodes are the structural constructs that shape the composition workflow. The second one is a graph-based approach where the atomic services are the vertices and the edges connecting them form the composition workflow. The two approaches are compared to determine which is most suitable to the QoS-aware fully automated Web service composition problem.  Secondly, this thesis proposes novel sequence-based approaches for Web service composition that use an indirect representation, i.e. they encode solutions as sequences of services. By representing solutions in this way, it is possible to initialise and evolve them without having to enforce their functional correctness. Then, before evaluating the fitness of each solution, a decoding algorithm is used to transform the sequence into the corresponding composition. The decoding algorithm builds the workflow using the ordering in the sequence as closely as possible when selecting the next service to be added, while at the same time generating a functionally correct structure.  Thirdly, this thesis treats Web service composition as a multi-objective problem, generating a set of trade-off solutions the user can choose from. More specifically, it proposes multi-objective approaches to fully automated Web service composition, which means that conflicting QoS attributes are independently optimised using a variety of representations that support flexible workflow structures. Additionally, a multi-objective and fully automated memetic approach that uses a local search operator to further improve the quality of solutions is proposed.  The following major contributions have been made in this thesis. Firstly, two approaches for Web service composition with direct representations were proposed. When the choice construct is not considered, the graph-based approach produces solutions of higher quality than those of the tree-based approach, but the opposite is true when the choice construct is included. Secondly, indirect representation approaches for Web service composition were proposed. These approaches perform well and can produce solutions with better quality than those found by the graph-based approach. Finally, we propose multi-objective approaches to fully automated service composition, employing different problem representations and a local search operator. The multi-objective approaches using the sequence-based representation were found to produce solutions with better overall quality.</p>

Sign in / Sign up

Export Citation Format

Share Document