To solve min–max optimization problems, decentralized adaptive methods have been presented over multi-agent networks. In the non-convex non-concave structure, however, existing decentralized adaptive min–max methods may be divergence due to the inconsistency in the adaptive learning rate. To address this issue, we propose a novel decentralized adaptive algorithm named DADAMC, where the consensus protocol is introduced to synchronize the adaptive learning rates of all agents. Furthermore, we rigorously analyze that DADAMC converges to an ϵ -stochastic first-order stationary point with O ( ϵ − 4 ) complexity. In addition, we also conduct experiments to verify the performance of DADAMC for solving a robust regression problem. The experimental results show that DADAMC outperforms state-of-the-art decentralized min–max algorithms.

Abstract The detection of pests and diseases in crops is currently a hot topic. The complexity of pest and disease object in the field, combined with inconsistent features across different levels, poses challenges for network detection. Additionally, the complex agricultural production environment tends to generate many interfering negative samples, which significantly complicates pest and disease differentiation. To address these two issues, the YOLOv7‐PSAFP network structure was first proposed. Based on YOLOV7, the progressive Spatial Adaptive Feature Pyramid (PSAFP) was introduced. Second, a combination of the Varifocal Loss and Loss Rank Mining loss functions was used for calculating the object loss, which reduces the interference of useless negative examples during training. On the filtered‐plant‐village‐dataset and rice‐corn pest dataset, the mAP results of YOLOv7‐PSAFP were 84.7 and 93.3, which are 2.9 and 2.1 higher than the baseline model (YOLOv7), respectively. The code for this paper is located at https://github.com/DuLJ72/PSAFP .

A pivotal problem in the Internet of Things (IoT) is resource allocation, where the goal is to optimize allocation strategies of IoT resources. In general, resource allocation problems are formulated as constrained optimization problems, which can be effectively solved by the zeroth-order Frank-Wolfe algorithm. However, the existing zeroth-order Frank-Wolfe algorithms suffer from slow convergence since they scale the zeroth-order gradient in all directions. For this reason, we propose a faster zeroth-order Frank-Wolfe algorithm, referred to as ZO-AdaSFW, which incorporates adaptive gradient methods and the variance-reduced technique (SPIDER) into the zeroth-order Frank-Wolfe algorithm. Moreover, ZO-AdaSFW can achieve the convergence rate of$O(T^{-1}) $in the convex setting, where T is the time horizon. Meanwhile, we also prove that ZO-AdaSFW has the best-known convergence rate$O(T^{-1/2})$among zeroth-order algorithms in the nonconvex setting. In addition, the experimental results show that the performance of ZO-AdaSFW outperforms state-of-the-art zeroth-order Frank-Wolfe algorithms on different applications.

Abstract Weed recognition is an inevitable problem in smart agriculture, and to realise efficient weed recognition, complex background, insufficient feature information, varying target sizes and overlapping crops and weeds are the main problems to be solved. To address these problems, the authors propose a real‐time semantic segmentation network based on a multi‐branch structure for recognising crops and weeds. First, a new backbone network for capturing feature information between crops and weeds of different sizes is constructed. Second, the authors propose a weight refinement fusion (WRF) module to enhance the feature extraction ability of crops and weeds and reduce the interference caused by the complex background. Finally, a Semantic Guided Fusion is devised to enhance the interaction of information between crops and weeds and reduce the interference caused by overlapping goals. The experimental results demonstrate that the proposed network can balance speed and accuracy. Specifically, the 0.713 Mean IoU (MIoU), 0.802 MIoU, 0.746 MIoU and 0.906 MIoU can be achieved on the sugar beet (BoniRob) dataset, synthetic BoniRob dataset, CWFID dataset and self‐labelled wheat dataset, respectively.

Functional signatures were allowed anyone to sign any messages in the range of function f , who possesses the secret key s k f . However, the existing construction does not satisfy the property of message and function privacy. In this paper, we propose a new notion which is called functional message authentication codes (MACs). In a functional MAC scheme, there are two types of secret keys. One is a master secret key which can be used to generate a valid tag for any messages. The other is authenticating keys for a function f , which can be used to authenticate any messages belonged to the range of f . Except the unforgeability, we require the proposed functional MAC to satisfy function and message privacy which indicates that the authenticating process reveals nothing other than the function values and the corresponding tags. We give a functional MAC construction based on a functional encryption (FE) scheme with function privacy, a perfectly binding commitment scheme, a standard signature scheme, and a symmetric encryption scheme with semantic security. Then, we show an application of functional MAC to constructing verifiable outsourcing computation, which ensures that the client does not accept an incorrect evaluation from the server with overwhelming probability.

Federated meta-learning (ML) is a promising optimization framework for the intelligent Internet of Things (IoT). However, the generalization ability of existing federated ML is limited because it is a bilayer structure, which has a more complex loss landscape. Moreover, the loss landscape of bilevel optimization has more saddle points and sharp points, which may lead to different generalization performances. Therefore, how to choose an optimal point is crucial for improving the generalization ability of federated ML. For this reason, this article proposes a provable federated ML algorithm by using the sharpness-aware minimization technique, referred to as FedAvg-sharp-MAML (FSM). Furthermore, we rigorously analyse the convergence and generalization bound of FSM. Specifically, when local iteration rounds$T=1$, the rate of$O(1/K)$can be achieved, where K is the number of global iterations. Furthermore, this rate can match the Per-Fedavg method. Meanwhile, we achieve a better generalization bound than the state of the art federated ML, where PAC-Bayesian generalization bounds are introduced in our analysis. Finally, we conduct some experiments to verify the performance of FSM. The experimental results show that the FSM has good generalization performance compared to the existing federated ML algorithms.

We introduce a novel notion of distributed functional signature. In such a signature scheme, the signing key for function f will be split into n shares sk f i and distributed to different parties. Given a message m and a share sk f i , one can compute locally and obtain a pair signature f i m , σ i . When given all of the signature pairs, everyone can recover the actual value f m and corresponding signature σ . When the number signature pairs are not enough, nobody can recover the signature f m , σ . We formalize the notion of function privacy in this new model which is not possible for the standard functional signature and give a construction from standard functional signature and function secret sharing based on one-way function and learning with error assumption. We then consider the problem of hosting services in multiple untrusted clouds, in which the verifiability and program privacy are considered. The verifiability requires that the returned results from the cloud can be checked. The program privacy requires that the evaluation procedure does not reveal the program for the untrusted cloud. We give a verifiable distributed secure cloud service scheme from distributed functional signature and prove the securities which include untrusted cloud security (program privacy and verifiability) and untrusted client security.

Abstract Functional encryption (FE) can provide a fine-grained access control on the encrypted message. Therefore, it has been applied widely in security business. The previous works about functional encryptions most focused on the deterministic functions. The randomized algorithm has wide application, such as securely encryption algorithms against chosen ciphertext attack, privacy-aware auditing. Based on this, FE for randomized functions was proposed. The existing constructions are provided in a weaker selective security model, where the adversary is forced to output the challenge message before the start of experiment. This security is not enough in some scenes. In this work, we present a novel construction for FE, which supports the randomized functionalities. We use the technology of key encapsulated mechanism to achieve adaptive security under the simulated environment, where the adversary is allowed to adaptively choose the challenge message at any point in time. Our construction is built based on indistinguishability obfuscation, non-interactive witness indistinguishable proofs and perfectly binding commitment scheme.

Constrained verifiable random functions (VRFs) were introduced by Fuchsbauer. In a constrained VRF, one can drive a constrained key skS from the master secret key sk , where S is a subset of the domain. Using the constrained key skS , one can compute function values at points which are not in the set S. The security of constrained VRFs requires that the VRFs’ output should be indistinguishable from a random value in the range. They showed how to construct constrained VRFs for the bit-fixing class and the circuit constrained class based on multilinear maps. Their construction can only achieve selective security where an attacker must declare which point he will attack at the beginning of experiment. In this work, we propose a novel construction for constrained verifiable random function from bilinear maps and prove that it satisfies a new security definition which is stronger than the selective security. We call it semiadaptive security where the attacker is allowed to make the evaluation queries before it outputs the challenge point. It can immediately get that if a scheme satisfied semiadaptive security, and it must satisfy selective security.

In this work, we first design a function encryption scheme by using key encapsulation. We combine public key encryption with symmetric encryption to implement the idea of key encapsulation. In the key encapsulation, we use a key to turn a message (plaintext) into a ciphertext by symmetric encryption, and then we use public key encryption to turn this key into another ciphertext. In the design of function encryption scheme, we use the public key encryption system, symmetric encryption system, noninteractive proof system, indistinguishable obfuscator, and commitment scheme. Finally, we prove the indistinguishable security of our function encryption scheme.