Neural architecture search (NAS) is an effective approach for automating the design of deep neural networks. Evolutionary computation (EC) is commonly used in NAS due to its global optimization capability. However, the evaluation phase of architecture candidates in EC-based NAS is compute-intensive, limiting its application for many real-world problems. To overcome this challenge, we propose a novel progressive evaluation strategy for the evaluation phase in convolutional neural network architecture search, in which the number of training epochs of network individuals is progressively increased. In addition, a subpopulation preservation strategy is proposed to preserve medium-size and large-size architectures to avoid prematurely discarding networks that may not perform well in the early stages but have the potential to excel with further optimization. Our proposed algorithm reduces the computational cost of the evaluation phase and promotes population diversity and fairness by preserving promising networks based on their distribution. We evaluate the proposed progressive evaluation and subpopulation preservation of NAS (PEPNAS) algorithm on the CIFAR10, CIFAR100, and ImageNet benchmark datasets, and compare it with 36 state-of-the-art algorithms, including manually designed networks, reinforcement learning (RL) algorithms, gradient-based algorithms, and other EC-based ones. The experimental results demonstrate that PEPNAS effectively identifies networks with competitive accuracy while also markedly improving the efficiency of the search process. For instance, PEPNAS discovers the architecture on CIFAR10 with a low-error rate of 2.38% using only 0.7 GPU days. We directly adopt the searched architecture for the image classification on the CIFAR100 and ImageNet datasets, which achieves the top 1 error rates of 16.46% and 26.25%, respectively. The code is available athttps://github.com/chajiajie/PEPNAS.

Neural architecture search (NAS) is a popular method that can automatically design deep neural network structures. However, designing a neural network using NAS is computationally expensive. This article proposes a gradient-guided evolutionary NAS (GENAS) to design convolutional neural networks (CNNs) for image classification. GENAS is a hybrid algorithm that combines evolutionary global and local search operators to evolve a population of subnets sampled from a supernet. Each candidate architecture is encoded as a table describing which operations are associated with the edges between nodes signifying feature maps. Besides, evolutionary optimization uses novel crossover and mutation operators to manipulate the subnets using the proposed tabular encoding. Every generations, the candidate architectures undergo a local search inspired by differentiable NAS. GENAS is designed to overcome the limitations of both evolutionary and gradient descent NAS. This algorithmic structure enables the performance assessment of the candidate architecture without retraining, thus limiting the NAS calculation time. Furthermore, subnet individuals are decoupled during evaluation to prevent strong coupling of operations in the supernet. The experimental results indicate that the searched structures achieve test errors of 2.45%, 16.86%, and 23.9% on CIFAR-10/100/ImageNet datasets and it costs only 0.26 GPU days on a graphic card. GENAS can effectively expedite the training and evaluation processes and obtain high-performance network structures.

Uncertainty modeling and reasoning in intelligent systems are crucial for effective decision-making, such as complex evidence theory (CET) being particularly promising in dynamic information processing. Within CET, the complex basic belief assignment (CBBA) can model uncertainty accurately, while the complex rule of combination can effectively reason uncertainty with multiple sources of information, reaching a consensus. However, determining CBBA, as the key component of CET, remains an open issue. To mitigate this issue, we propose a novel method for generating CBBA using high-level features extracted from Box–Cox transformation and discrete Fourier transform (DFT). Specifically, our method deploys complex Gaussian fuzzy number (CGFN) to generate CBBA, which provides a more accurate representation of information. The proposed method is applied to pattern classification tasks through a multisource information fusion algorithm, and it is compared with several well-known methods to demonstrate its effectiveness. Experimental results indicate that our proposed CGFN-based method outperforms existing methods, by achieving the highest average classification rate in multisource information fusion for pattern classification tasks. We found the Box–Cox transformation contributes significantly to CGFN by formatting data in a normal distribution, and DFT can effectively extract high-level features. Our method offers a practical approach for generating CBBA in CET, precisely representing uncertainty and enhancing decision-making in uncertain scenarios.

The law of total probability plays an essential role in Bayesian reasoning, which has been used in many fields. However, some experiments show the law of total probability can be violated. In recent years, researchers have tried to explain this paradox with the interference effect in quantum theory, and they think the main reason for interference effects is the uncertain information in the decision-making process. Therefore, how to effectively model and process the uncertain information in the decision-making process is very important to understand and predict the interference effects. Zadeh proposed the fuzzy set by considering the fuzziness of information. Later, Atanassov proposed the intuitionistic fuzzy sets (IFS). IFS better describes the fuzzy information from the view of membership, nonmembership than fuzzy sets, which can also more flexibly simulate human decision making. Hence, the article proposed the fuzzy Markov decision-making model (FDM) under the framework of IFS to explain and predict the interference effects of decision-making process. In FDM, intuitionistic fuzzy number can be generated by using the negation operation of probability. In addition, the transition matrix can be obtained by using the Kolmogorov equation, which can consider the evolution time in the decision-making process. The transition matrix establishes the relationship between different stages to get the fuzzy numbers of final states. Finally, the article used the Dempster–Shafer evidence theory to transform fuzzy number into the probability. In summary, the proposed FDM can provide a novel idea to explore and explain the interference effects in the decision-making process, which is helpful to promote the development of artificial intelligence.

This 21st century is notable for experiencing so many disturbances at economic, social, cultural, and political levels in the entire world. The outbreak of novel corona virus 2019 (COVID-19) has been treated as a Public Health crisis of global Concern by the World Health Organization (WHO). Various outbreak models for COVID-19 are being utilized by researchers throughout the world to get well-versed decisions and impose significant control measures. Amid the standard methods for COVID-19 worldwide epidemic prediction, easy statistical, as well as epidemiological methods have got more consideration by researchers and authorities. One main difficulty in controlling the spreading of COVID-19 is the inadequacy and lack of medical tests for detecting as well as identifying a solution. To solve this problem, a few statistical-based advances are being enhanced and turn into a partial resolution up-to some level. To deal with the challenges of the medical field, a broad range of intelligent based methods, frameworks, and equipment have been recommended by Machine Learning (ML) and Deep Learning. As ML and DL have the ability of identifying and predicting patterns in complex large datasets, they are recognized as a suitable procedure for producing effective solutions for the diagnosis of COVID-19. In this paper, a perspective research has been conducted in the applicability of intelligent systems such as ML, DL and others in solving COVID-19 related outbreak issues. The main intention behind this study is (i) to understand the importance of intelligent approaches such as ML and DL for COVID-19 pandemic, (ii) discussing the efficiency and impact of these methods in the prognosis of COVID-19, (iii) the growth in the development of type of ML and advanced ML methods for COVID-19 prognosis,(iv) analyzing the impact of data types and the nature of data along with challenges in processing the data for COVID-19,(v) to focus on some future challenges in COVID-19 prognosis to inspire the researchers for innovating and enhancing their knowledge and research on other impacted sectors due to COVID-19.

Analysis and modeling of the intrusion detection system is an important phenomenon for any communication network, which helps to monitor the network traffic and avoid suspicious activity in the Big Data environment. The machine learning approach for modeling the intrusion detection system requires analysis of large network data, which may include some irrelevant features resulting in unnecessary computational and analytical burden. In this article, a fuzzy and real coded chemical reaction optimization-based cluster analysis approach with feature selection is proposed for the intrusion detection system in a Big Data platform. The proposed cluster analysis model is achieved through Fuzzy C-Mean (FCM) with real-coded chemical reaction optimization, which boosts FCM to start with optimized cluster centers. Also, the use of the Flexible Mutual Information Feature Selection approach helps this model to avoid the processing of a large number of features, which drastically affects processing elements.

Salp Swarm Algorithm (SSA) is a novel swarm technique using to optimize design problems. SSA is inspired by the swarming behavior of salp observed in the deep area of sea. In spite of its application versatility, SSA suffers from mediocre convergence rate and limited exploratory capabilities. In this paper, a Fitness Dependent Salp Swarm Algorithm (FDSSA) is proposed. The novelties of the proposed approach are the definition of a fitness coefficient able to enhance the exploration, the introduction of a novel mathematical model to describe the trajectory of salps and a mutation mechanism to increase the convergence speed. The designed algorithm is tested on unimodal and multimodal benchmark functions and then compared with well-known heuristic algorithms. The results show the superiority of FDSSA with respect to the comparison algorithms in terms of optimization and convergence performances according to their computational complexity.

Internet of Things (IoT) has evolved for building smart environments in a distributed system, where the data produced by IoT devices are transmitted through Edge computing devices to streamline the flow of traffic from IoT devices to a distributed network. In such a scenario, the attacker introduces many attacks to the edge before forwarding them to distributed servers. This necessitates intrusion detection systems for such environments to mitigate security attacks. This paper has projected a basis for characterization of intrusive behaviors in a distributed system based on the functional link neural nets response weighted-average and teaching-learning metaheuristic with elitism on weight-space. The proposed technique makes use of teaching-learning metaheuristic optimization to obtain suitable parameters for the functional link neural net. Furthermore, the processing of duplicate parameters is successfully avoided by using mutation operation. In addition to this, in this paper the proposed method is found to be more efficient in terms of computational burden.

A new efficient design of second-order spectralnull (2-OSN) codes is presented. The new codes are obtained by applying the technique used to design parallel decoding balanced (i.e., 1-OSN) codes to the random walk method introduced by some of the authors for designing 2-OSN codes. This gives new non-recursive efficient code designs, which are less redundant than the code designs found in the literature. In particular, if k ∈ IIN is the length of a 1-OSN code then the new 2-OSN coding scheme has length n = k + r ∈ IIN with an extra redundancy of r ≃ 2 log2k + (1/2) log2log2k - 0.174 check bits, with k and r even and n multiple of 4. The whole coding process requires O(k log k) bit operations and O(k) bit memory elements.

An m-ary block code, m = 2, 3, 4,..., of length n ϵ IN is called balanced if, and only if, every codeword is balanced; that is, the real sum of the codeword components, or weight, is equal to ⌊(m - 1)n/2⌋. This paper presents efficient encoding schemes to m-ary balanced codes with parallel (hence, fast) decoding. In fact, the decoding time complexity is O(1) digit operations. These schemes are a generalization to the m-ary alphabet of Knuth's complementation method with parallel decoding. Let (nw)mindicate the number of m-ary words w of length n and weight w ϵ(0,1, ... , (m - 1)n}. For any m ϵ IN, m ≥ 2, a simple implementation of the method is given which uses r ϵ IN check digits to balance k ≤ {(⌊(m-1)r/2⌋)m- (m mod 2 + [(m - 1)k] mod 2}}/(m - 1) information digits with an encoding time complexity of O(mk logmk) digit operations. A refined implementation of the parallel decoding method is also given with r check digits and k ≤ (mr-1)/(m -1) information digits, where the encoding time complexity is O(k√logmk). Thus, the proposed codes are less redundant than the m-ary balanced codes with parallel decoding found in the literature and yet maintain the same complexity.

A new efficient coding scheme is given for second-order spectral-null (2-OSN) codes. The new method applies the Knuth's optimal parallel decoding scheme for balanced (i.e., 1-OSN) codes to the random walk method introduced by Tallini and Bose to design 2-OSN codes. If k ∈ IN is the length of a 1-OSN code then the new 2-OSN coding scheme has length n = k+r ∈ IN with an extra redundancy of r ≳ 2 log2k + (1/2) log2log2k - 0.674 check bits. The whole coding process requires O(n log n) bit operations and 0(n) bit memory elements.

An m-ary block code, m = 2, 3, 4, ..., of length n ∈ IN is called balanced if, and only if, every codeword is balanced; that is, the real sum of the codeword components, or weight, is equal to ⌊(m - 1)n/2⌋. This paper presents a tight generalization of Knuth's complementation method with parallel (hence, fast) decoding scheme. Let (wn)mindicate the number of m-ary words of length n and weight w ∈ {0, 1, ..., (m-1)n}. A simple implementation of the scheme uses (m - 1)k + m mod 2 balancing functions to make a k ∈ IN digit information word to be balanced. So, r ∈ IN check digits can be used to balance k ≤ [(⌊(m-1)rr/2⌋)m-m mod 2]/(m - 1) information digits. A refined implementation of the parallel decoding scheme uses r check digits to balance k ≤ (mr-1)/(m-1) information digits.