Research Library
The literature behind our research
Recent Q1–Q2 journal articles connected to DeepMathAI's research, with a short summary of each and links to the publisher and to free versions.
Popular searches:
Neurodynamic Optimization
71 articlesRecurrent and gradient neural networks whose dynamics solve optimization, equilibrium and matrix-equation problems — with fixed- and predefined-time convergence, distributed and collaborative variants, and applications to machine-learning models.
DeepMathAI work in this area
Gradient neural networks11
- New2026A fully quaternion-valued Adams–Moulton discrete-time integration-enhanced zeroing neural network for solving time-varying quaternion equations
Jun Ma, Jianle Peng, Pei Liu et al.
Read the abstract on the publisher's site - New2026A modified artificial potential field approach for real-time robot path planning via adaptive gradient neural network
Zisheng Chen, Ting Li, Ying Liufu
TL;DR
A modified artificial potential field approach based on adaptive gradient neural network (AGNN) presented to generate safe and smooth real-time paths and theoretical guarantees of global asymptotic convergence and obstacle impenetrability are rigorously established.
- New2026An adaptive anti-noise zeroing neural network for solving time-varying linear systems and its applications
Dongmei Yu, Hongrui Yin, Shaowei Lin et al.
Overview
Designs an adaptive zeroing neural network that stays accurate under noise when solving time-varying linear systems, and applies it to example problems.
- New2026Discrete-time zeroing neural networks: Models, developments, and robotic applications—A survey
Angjin Li, Zhijun Zhang, Zhuosong Fu
Read the abstract on the publisher's site - 2026Error orthogonalization gradient neural network: A novel discrete paradigm for time-varying linear equations
Shangfeng Du, Yang Si, Yafeng Zhong et al.
Overview
Introduces a discrete-time gradient neural network built on error orthogonalization, a new paradigm for tracking the solutions of time-varying linear equations.
- New2026Multicriteria model predictive control of redundant manipulators using a null-space active-screening predictor-corrector discrete zeroing neural network
Chao Chao Zhong, Yu Han, Binbin Qiu
Overview
Combines multicriteria model predictive control with a discrete zeroing neural network that uses null-space active screening and a predictor–corrector scheme to control redundant robot manipulators.
- New2026Unifying the Moore-Penrose conditions: A non-singular zeroing neural network with neutrosophic self-adaptive integration
Rubayyi T. Alqahtani, Nadiyah Hussain Alharthi, Theodore E. Simos et al.
Read the abstract on the publisher's site - 2025First/second-order norm-based gradient neural networks for dynamic linear matrix equation
Chunhao Han, Jiao Hui Xu
Overview
Proposes first- and second-order gradient neural networks driven by norm-based error functions for solving dynamic (time-varying) linear matrix equations.
- 2025Gradient neural network based solutions to periodic coupled sylvester equations
Lingling Lv, Jiali Zhao, Jinbo Chen et al.
Overview
Develops gradient neural network solvers for periodic coupled Sylvester matrix equations, which arise in the analysis and control of periodic systems.
- 2024Gradient neural network model for the system of two linear matrix equations and applications
Jelena Dakić, Marko D. Petković
Overview
Constructs a gradient neural network model that solves a system of two linear matrix equations simultaneously, and presents applications of the model.
- 2024Predefined-time robust adaptive gradient neural network for solving linear time-varying equations and its applications
Dongmei Yu, Gehao Zhang, Tong Zhang
Overview
Presents an adaptive gradient neural network that is robust to disturbances and converges within a predefined time when solving linear time-varying equations, with applications.
Primal–dual neural networks5
- 2026Augmented Lagrangian Neural Network for Solving Mathematical Programs with Equilibrium Constraints
Anjali Rawat, Vinay Singh
TL;DR
It is demonstrated that the neural network converges to an equilibrium point that satisfies the Karush-Kuhn-Tucker conditions of the underlying nonlinear programming problem, guaranteeing the optimality of the obtained solution.
- 2026Learning convex set boundaries via primal-dual neural approximation with application to reachable set computation
Guopeng Chen, Lizhen Shao, Fangyuan Zhao
TL;DR
A learning-based framework, the Kolmogorov-Arnold Network enhanced Primal-Dual Neural Network (KAN-PDNN), which approximates the solution map of parameterized optimization problems that implicitly define such sets.
- 2025A noniterative linear-variational-inequality based primal-dual neural network for repetitive motion planning of robots
Weibing Li, Ruiqi Rao, Yuxuan Chen et al.
Overview
Formulates repetitive motion planning of robot manipulators as a linear variational inequality and solves it with a noniterative primal–dual neural network.
- 2025Performing task automation for surgical robot: A spatial–temporal varying primal–dual neural network with guided obstacle avoidance and null space optimization
Xingqiang Jian, Bo Wu, Yibin Song et al.
From the abstract
Performing surgical tasks safely and reliably presents significant challenges, including obstacle avoidance, joint limit constraints, and motion smoothness during the tool-target alignment (T-TA) stage, as well as precise tracking of preoperative plans during the execution of the preoperative planning surgery path (EPSP).
- 2024Integrated guidance and control based on nonlinear model predictive control with simplified primal–dual neural network optimization
Lei Cui, Zhe Zhen, Kang Zhang et al.
Overview
Integrates guidance and control through nonlinear model predictive control, solving the underlying optimization online with a simplified primal–dual neural network.
Projection neural networks9
- New2026A relaxed inertial projection algorithm for solving inverse quasi-variational inequalities and its application to traffic assignment
Konrawut Khammahawong, Jamilu Abubakar, Abubakar Adamu
From the abstract
In this article, we propose a relaxed inertial projection algorithm for solving inverse quasi-variational inequality problems (IQVIPs) in real Hilbert spaces. The method integrates inertial extrapolation and relaxation techniques to accelerate the convergence of the classical projection scheme.
- New2026An implicit-curvature projection neural network for nonlinear programming
Yang Si, Zhibin Li, Kai Zhao et al.
TL;DR
This article proposes an implicit-curvature projection neural network (ICPNN) for nonlinear programming (NLP) that contains a slow decision layer and a coordinated fast correction layer that induces a curvature-sensitive structural effect without explicitly computing second-order information.
- 2026An inertial projection neural network for solving inverse quasi-variational inequalities
Qun Li, Guolin Chen, Maolin Liang
TL;DR
It is shown that the proposed IPNN method can guarantee the uniqueness of the solution to the considered IQVI problem under appropriate conditions and is not only asymptotically stable but also exponentially stable at the equilibrium solution.
- 2026Scaled projection neural network for quasi-variational inequalities
Mohammed Alshahrani, Qamrul Hasan Ansari
From the abstract
We propose a scaled projection neural network for solving quasi-variational inequalities in which the constraint set S(x)=m(x)+S depends on the state through a contraction mapping m that translates a fixed closed convex set S, while the projection is performed with respect to a fixed symmetric positive definite matrix M.
- New2026Time-varying momentum-like proximal neurodynamics with fixed-time convergence for non-smooth nash equilibrium seeking
Hao Li, Abdujelil Abdurahman, Haijun Jiang et al.
Overview
Proposes proximal neurodynamics with a time-varying, momentum-like term that converge in fixed time to Nash equilibria of non-smooth games.
- 2024A proximal neurodynamic model for a system of non-linear inverse mixed variational inequalities
Anjali Upadhyay, Rahul Pandey
TL;DR
The uniqueness of the continuous solution for the PNDM is proved by assuming Lipschitz continuity and the global asymptotic stability of equilibrium points of the PNDM is established, contingent upon Lipschitz continuity and strong monotonicity.
- 2024An efficient and robust varying-parameter projection neural network for sparse signal reconstruction
Qing He Hu, Bing Zheng
Overview
Develops a projection neural network with varying parameters for sparse signal reconstruction, designed for efficient and robust recovery.
- 2024Impulsive Projection Neural Networks for Variational Inequalities and Sparse Signal Reconstruction Application
Jing Xu, Chuandong Li, Xing He et al.
TL;DR
The newly designed IPNN and MIPNN have an improvement in the convergence rates because of the introduction of impulsive effects and the effectiveness and feasibility of the IPNN and MIPNN are shown.
- New2020The forward–backward–forward method from continuous and discrete perspective for pseudo-monotone variational inequalities in Hilbert spaces
Radu Ioan Boţ, Ernö Robert Csetnek, Phan Tu Vuong
TL;DR
This paper associates to a pseudo-monotone variational inequality a forward-backward-forward dynamical system and carries out an asymptotic analysis for the generated trajectories and proves that linear convergence is guaranteed under strong pseudo- monotonicity.
Fixed- and predefined-time convergence6
- New2026Predefined-time zeroing neural network for solving the Yang-Baxter-like matrix equation
Yiqiong Sun, Qing-Wen Wang, Duanmei Zhou
Overview
Designs a zeroing neural network that converges within a predefined time to solutions of the Yang–Baxter-like matrix equation AXA = XAX.
- 2026Second-order predefined-time neural networks for sparse recovery
Yuan-Min Li, Ke Zheng
Overview
Introduces neural networks with second-order dynamics that converge within a predefined time for sparse recovery problems.
- 2025A predefined-time matrix-value neural network for quaternion optimization with applications to color images
Banghua Huang, Yang Liu, Kit Ian Kou et al.
Overview
Proposes a matrix-valued neural network with predefined-time convergence for optimization over quaternions, applied to colour images.
- 2025Predefined-time with time-varying coefficients neurodynamic for composite optimization problems
Dongmei Yu, Shaowei Lin, Gehao Zhang et al.
Overview
Develops neurodynamic approaches with time-varying coefficients that solve composite optimization problems within a predefined time.
- 2024A novel predefined-time neurodynamic approach for mixed variational inequality problems and applications
Jinlan Zheng, Xingxing Ju, Naimin Zhang et al.
TL;DR
This paper proposes a novel neurodynamic approach with predefined-time stability that offers a solution to address mixed variational inequality problems and introduces an adjustable time parameter, thereby enhancing flexibility and applicability compared to conventional fixed-time stability methods.
- 2024New results on fixed-time synchronization of impulsive neural networks via optimized fixed-time stability
Abdujelil Abdurahman, Rukeya Tohti, Cuicui Li
TL;DR
The impulsive gains in this paper not only can be stabilizing but also can be destabilizing, which provides wider application scope for the finite-time and fixed-time stability and synchronization studies of other impulse systems.
Distributed and multi-agent optimization7
- New2026Distributed double-layer neurodynamic optimization for dynamic constrained system with disturbance rejection
Jingxin Liu, Lijun Wang, Jingwen Tu et al.
Overview
Proposes a distributed double-layer neurodynamic scheme for constrained optimization of dynamic systems that also rejects external disturbances.
- New2026Distributed neurodynamic optimization for multi-UAV task allocation with communication constraints and convergence guarantees
Wang Xin, Shuangfeng Ma
Overview
Applies distributed neurodynamic optimization to task allocation among multiple UAVs under communication constraints, with convergence guarantees.
- 2025A reformulation neurodynamic algorithm for distributed nonconvex optimization
Xin Ye Yu, Qingzhou Huang, Rixin Lin
Overview
Presents a neurodynamic algorithm based on reformulating the problem, for distributed nonconvex optimization over networks of agents.
- 2025Distributed neurodynamic optimization for optimal multicluster resource allocation with cardinality constraints
Meng Hsien Lin, Zicong Xia, Shuting Sun et al.
Overview
Uses distributed neurodynamic optimization to allocate resources optimally across multiple clusters subject to cardinality constraints.
- 2024A collective neurodynamic approach to distributed resource allocation with event-triggered communication
Xin Cai, Bingpeng Gao, Xinyuan Nan
TL;DR
It is analyzed that the convergence of the designed collective neurodynamic approach based on the event-triggered communication does not rely on global information, and it is proved the freeness of the Zeno behavior in the event-triggered scheme.
- 2024A p-power neurodynamic approach to distributed nonconvex optimization
Yangxia Li, Zicong Xia, Yang Liu et al.
Overview
Proposes a p-power neurodynamic approach for distributed nonconvex optimization over networked agents.
- 2024Distributed nonconvex optimization subject to globally coupled constraints via collaborative neurodynamic optimization
Zicong Xia, Yang Liu, Cheng Hu et al.
TL;DR
A recurrent neural network is proposed for distributed nonconvex optimization subject to globally coupled (in)equality constraints and local bound constraints and a collaborative neurodynamic optimization method is established by utilizing multiple proposed recurrent neural networks and a meta-heuristic rule.
Complementarity and merit functions10
- New2026Ellipsoidal cone eigenvalue complementarity problems: Solvability conditions, reformulations and numerical experiments
Yue Lu, Miao Liu, Ming Huang et al.
Read the abstract on the publisher's site - New2026On the lower and upper bounds of the solution set of the tensor complementarity problem
Jun He, Shi-Liang Wu, Yan-Min Liu
From the abstract
The tensor complementarity problem (TCP) is to find a vector x∈Rn such that x≥0,q+Axm−1≥0,x⊤(Axm−1+q)=0 with A∈R[m,n] and q∈Rn. In this paper, we further investigate the lower and upper bounds of the solution set of the tensor complementarity problem with q∈R+n.
- New2026Projected splitting methods for solving the generalized vertical polynomial complementarity problem
Dan Wang, Ji-Cheng Li
Read the abstract on the publisher's site - New2026Some modulus-based iteration methods for complex nonlinear complementarity problems
Zhiyuan Tong, Hua Zheng, Fan Bu et al.
Read the abstract on the publisher's site - New2026Sparse Solutions to Stochastic Tensor Complementarity Problem
Guangxuan Lin, Shouqiang Du
TL;DR
A regularized formulation of the approximate model is presented and the connection between the solutions of the two models is also established, and a gradient projection algorithm is designed for this regularized model and establishing its convergence result.
- 2025Polyhedral Newton-min algorithms for complementarity problems
Jean‐Pierre Dussault, Mathieu Frappier, Jean Charles Gilbert
TL;DR
A globally convergent algorithm is proposed using a modification of a semismooth Newton direction that makes it a descent direction of the least-squares function, instead of requiring that the direction satisfies a linear system; it must be a feasible point of a convex polyhedron; hence, it can be computed in polynomial time.
- 2024A least-change secant algorithm for solving generalized complementarity problem
Hevert Vivas, Rosana Pérez M, C. Arias
From the abstract
In this paper, we propose a least-change secant algorithm to solve the generalized complementarity problem indirectly trough its reformulation as a nonsmooth system of nonlinear equations using a one-parametric family of complementarity functions.
- 2024A Neural Network Approach for Solving Weighted Nonlinear Complementarity Problems
Shuilian Xie, Zhen-Ping Yang, Hongru Xu
TL;DR
This paper reformulates the weighted nonlinear complementarity problem as an unconstrained minimization problem and presents the steepest descent-based neural network to solve it, and derives some results related to the asymptotic stability of the neural network.
- 2024Delayed neural network based on a new complementarity function for the NCP
Yuan-Min Li, Tianyv Lei
Overview
Builds a delayed neural network on a new complementarity function to solve the nonlinear complementarity problem (NCP).
- 2024New smooth weighted complementarity functions and a cubically convergent method for wLCP
Tiantian Fan, Jingyong Tang
From the abstract
The weighted linear complementarity problem (wLCP) can be used for modelling a large class of problems from science and economics. In this paper, we introduce a new class of weighted complementarity functions and show that it is continuously differentiable everywhere.
Neurodynamics for machine-learning models17
- 2026Fixed-time neurodynamic algorithms with element-wise normalization for sparse signal recovery
Hongsong Wen, Xing He, You Zhao et al.
Overview
Designs fixed-time neurodynamic algorithms with element-wise normalization for recovering sparse signals.
- 2026Neurodynamical neural network method for solving support vector machines with epigraph reformulated generalized pinball loss function
Rabian Wangkeeree, Rattanaporn Wangkeeree, Yirga Abebe Belay et al.
Overview
Rewrites support vector machines with the generalized pinball loss in epigraph form and solves them with a neurodynamical neural network, a continuous-time solver for robust SVM classification.
- 2025Neurodynamic optimization algorithm for split equality problems and application to sparse signal recovery
Binghao OuYang, Yong Wang, Xingxing Ju et al.
Overview
Proposes a neurodynamic optimization algorithm for split equality problems and applies it to sparse signal recovery.
- 2024A second-order projection neurodynamic approach with exponential convergence for sparse signal reconstruction
Chunhao Han, Jiao Hui Xu, Bing Zheng
Overview
Develops a second-order projection neurodynamic approach with exponential convergence for sparse signal reconstruction.
- 2024Time-varying neurodynamic optimization approaches with fixed-time convergence for sparse signal reconstruction
Xingxing Ju, Xinsong Yang, Linbo Qing et al.
Overview
Presents time-varying neurodynamic optimization approaches that reconstruct sparse signals with fixed-time convergence.
Collaborative neurodynamic optimization in machine learning
- 2027A survey on collaborative neurodynamic optimization
Jun An Yang, Xiao Lin, Donghui Zhang et al.
Overview
A survey of collaborative neurodynamic optimization, in which several neurodynamic models work together to search for global optima of nonconvex and combinatorial problems.
- 2026A survey on neurodynamics of deep learning: A unified perspective on architectures and optimization algorithms
Tao Shi, Liangming Chen, Zhengtai Xie et al.
Overview
A survey that views deep-learning architectures and their training algorithms through neurodynamics, offering a unified perspective on both.
- 2024Binary matrix factorization via collaborative neurodynamic optimization
Hongzong Li, Jun Wang, Nian Zhang et al.
TL;DR
This paper presents a collaborative neurodynamic optimization approach to binary matrix factorization based on the original combinatorial optimization problem formulation and quadratic unconstrained binary optimization problem reformulations.
- 2024Solving Constrained Pseudoconvex Optimization Problems with deep learning-based neurodynamic optimization
Dawen Wu, Abdel Lisser
Overview
Solves constrained pseudoconvex optimization problems by combining neurodynamic optimization with deep learning.
- 2022Balanced clustering based on collaborative neurodynamic optimization
Xiangguang Dai, Jun Wang, Wei Zhang
Overview
Formulates balanced clustering as an optimization problem and solves it with collaborative neurodynamic optimization.
- 2022Boolean matrix factorization based on collaborative neurodynamic optimization with Boltzmann machines
Xinqi Li, Jun Wang, Sam Kwong
TL;DR
This paper presents a collaborative neurodynamic approach to Boolean matrix factorization based on a binary optimization formulation to minimize the Hamming distance between a given data matrix and its low-rank reconstruction.
- 2021A neurodynamic optimization approach to supervised feature selection via fractional programming
Yadi Wang, Xiaoping Li, Jun Wang
TL;DR
Experimental results with eight benchmark datasets are discussed to demonstrate the global convergence of the neural networks and superiority of the proposed neurodynamic approach to several existing feature selection methods in terms of classification accuracy, precision, recall, and F-measure.
- 2021Two-timescale neurodynamic approaches to supervised feature selection based on alternative problem formulations
Yadi Wang, Jun Wang, Hangjun Che
TL;DR
The fractional programming problem formulation is equivalently reformulated as bilevel and bilinear programming problems without using any fractional function to obviate the fractional complexity.
Sparse recovery and low-rank problems
- 2025Inertial primal–dual projection neurodynamic approaches for constrained convex optimization problems and application to sparse recovery
You Zhao, Zhihua Allen-Zhao, Lei Wang et al.
TL;DR
This paper presents a centralized primal - dual projection neurodynamic approach with time scaling (CPDPNA-TS), built upon the heavy - ball method, which contains a second-order projection ODE with derivative feedback for the primal variables and a first-order ODE for the dual variables.
- 2024A continuous-time neurodynamic approach in matrix form for rank minimization
Meng Zhang, Xing He
TL;DR
The effectiveness of the proposed continuous-time neurodynamic approach for solving the rank minimization under affine constraints has been demonstrated, along with its superiority compared to the vector-based approach.
- 2024Distributed continuous-time accelerated neurodynamic approaches for sparse recovery via smooth approximation to L1 -minimization
Junpeng Xu, Xing He
TL;DR
This paper develops two continuous-time distributed accelerated neurodynamic approaches for solving sparse recovery via smooth approximation to L1-norm minimization problem and proves that the two proposed distributed neurodynamic approaches both achieve O(1t2) convergence rate.
- 2022Neurodynamic approaches for sparse recovery problem with linear inequality constraints
Jiao Yang, Xing He, Tingwen Huang
TL;DR
Two neurodynamic approaches for solving the L1-minimization problem with the linear inequality constraints are developed and relevant theories show that each agent globally converges to an optimal solution of the distributed optimization problem.
Inertial dynamics and ODE methods6
- New2026From inertial dynamics with implicit Hessian damping to accelerated forward–backward algorithms
Xin He, Ya-Ping Fang
Overview
Starts from inertial dynamics with implicit Hessian damping and derives accelerated forward–backward algorithms from them.
- 2026Heavy Ball and Nesterov Accelerations with Hessian-Driven Damping for Nonconvex Optimization
Nicolas Hadjisavvas, Felipe Lara, R. T. Marcavillaca et al.
TL;DR
A second-order dynamical system with Hessian-driven damping tailored for a class of nonconvex functions called strongly quasiconvex is investigated, and two discrete-time gradient-based algorithms are derived through time discretizations to enhance stability and convergence performance.
- 2026Second-Order Inertial Dynamics with Viscous and Hessian-Driven Damping for Linearly Constrained Convex Optimization
Chuanlong Xu, Shouqiang Du, Cheng Xu et al.
From the abstract
This paper studies a second-order inertial dynamical system combining asymptotically vanishing viscous damping with Hessian-driven damping for linearly constrained convex optimization.
- 2024Accelerated optimization through time-scale analysis of inertial dynamics with asymptotic vanishing and Hessian-driven dampings
Samir Adly, Hédy Attouch
TL;DR
This study focuses on the rapid optimization properties of inertial dynamics, which combine asymptotically vanishing viscous damping with Hessian-driven damping, and examines the case where the limit dynamic corresponds to the Levenberg-Marquardt regularization of Newton's continuous method.
- 2024Fast Convex Optimization via Differential Equation with Hessian-Driven Damping and Tikhonov Regularization
Gangfan Zhong, Xiaozhe Hu, Ming Qiao Tang et al.
TL;DR
By constructing proper energy function, it is proved that the trajectory strongly converges to a minimizer of the objective function of minimum norm, and a gradient-based optimization algorithm based on numerical discretization is proposed, and demonstrated its effectiveness in numerical experiments.
- 2024Solving convex optimization problems via a second order dynamical system with implicit Hessian damping and Tikhonov regularization
Szilárd Csaba László
From the abstract
This paper deals with a second order dynamical system with a Tikhonov regularization term in connection to the minimization problem of a convex Fréchet differentiable function.
Optimization Theory
42 articlesCone programming, absolute value equations, smoothing of nonsmooth problems, multiobjective and semidefinite optimization, and the optimality conditions behind them.
DeepMathAI work in this area
Second-order and circular cones6
- DeepMathAI2026A projection recurrent neural network method for solving absolute value equations associated with second-order cones
Rattanaporn Wangkeeree, Rabian Wangkeeree, Alireza Nazemi et al.
Overview
Proposes a projection recurrent neural network for absolute value equations associated with second-order cones and proves its global convergence to the exact solution using Lyapunov stability theory.
- New2026A spatial global optimization algorithm for generalized fractional programming problems using second-order cone relaxation
Zhisong Hou, Zhenwei Zhao, Deke Guo et al.
Overview
Develops a spatial global optimization algorithm for generalized fractional programming problems, built on second-order cone relaxation.
- 2026Smooth-like lower order penalty approach for solving second-order cone mixed complementarity problems
Zijun Hao, Qiong Wu, Chunmin Zhao
Overview
Proposes a smooth-like lower-order penalty approach for solving second-order cone mixed complementarity problems.
- 2026Smoothing methods for mathematical programs with second-order cone complementarity constraints
Yan-Chao Liang, En-Hui Xia, Gui-Hua Lin et al.
From the abstract
This paper studies mathematical programs with second-order cone complementarity constraints (SOCMPCCs), which generalize classical mathematical programs with complementarity constraints by incorporating second-order cone structures.
- 2026Weak Constraint Qualifications for Mathematical Programs with Second-Order Cone Complementarity Constraints
Yan-Chao Liang, Zhu Chen-yuan, Gui-Hua Lin et al.
From the abstract
This paper studies mathematical programs with second-order cone complementarity constraints (SOCMPCCs), which generalize classical mathematical programs with complementarity constraints.
- 2025A new smoothing homotopy algorithm for second-order cone problems
Tingting Xu, Xiaona Fan, Zaiyong Feng
From the abstract
In this paper, we introduce a new class of CHKS smoothing functions and show that the new smoothing functions possess a system of desirable properties. Based on the new smoothing functions, we obtain a smoothing homotopy method for solving the second-order cone complementarity problem.
Absolute value equations9
- New2026A globally convergent hybrid-adaptive accelerated Douglas–Rachford splitting method for absolute value equations
Pengjie Liu, Linhao Li, Peng Huang et al.
From the abstract
In this paper, we propose a hybrid-adaptive accelerated Douglas–Rachford splitting method (HAA-DRSM) for solving absolute value equations of the form Ex−|x|−p=0.
- New2026A new class of stochastic absolute value equations
Jingyong Tang, Rongge Chen
From the abstract
This paper investigates a new class of stochastic absolute value equations (SAVE), which is equivalent to the stochastic vertical linear complementarity problem. Due to the existence of random variables, the SAVE may not have a common solution for almost every realization.
- New2026More on the properties of solution set of absolute value equations
Shweta Yadav, Dipti Dubey
Read the abstract on the publisher's site - New2026On the Kaczmarz method with maximal residual rule for system of absolute value equations
Jiayi Wang, Fang Chen
Overview
Studies a Kaczmarz-type iterative method with a maximal-residual selection rule for solving systems of absolute value equations.
- 2025Absolute value equations with 2ⁿ solutions
Milan Hladík
From the abstract
An absolute value equation system is an algebraic problem that involves solving the n-by-n system Ax+|x|=b . This problem is known to be NP-hard, and its solution set can have a complicated structure. So far, the research has primarily focused on the unique solvability cases.
- 2025Absolute value equations with interval uncertainty
Milan Hladík, Lenka Ptáčková
From the abstract
We consider the (generalized) absolute value equations with uncertain entries. In particular, we assume that the entries come from given intervals. For such interval-valued problem, we study the fundamental properties regarding solutions and solvability.
- 2025An application of generalized absolute value equation
Cui-Xia Li, Shi-Liang Wu
From the abstract
In this paper, an application of the generalized absolute value equation is exploited. Concretely, by using the generalized absolute value equation, we provide a new approach to efficiently obtain the hull of the solution set to interval linear equations with certain inverse positive coefficient interval matrix.
- 2025Fractional absolute value equations
Francesco Mezzadri
From the abstract
We introduce the fractional absolute value equation (FAVE), which is an absolute value equation containing fractional terms of the unknown variables. For simplicity, we first consider a simple 1-dimensional FAVE and then generalize the results to problems of arbitrary dimension n.
- 2025Sparse Solutions to Tensor Absolute Value Equations
Shouqiang Du, Jingjing Sun, Yimin Wei
From the abstract
Sparse optimization problems have drawn wide attention in recent years, which have been extensively applied to compressed sensing, signal and image processings. As a kind of important optimization problem, absolute value equations have been extensively studied.
Smoothing and nonsmooth optimization11
- New2026A novel modiffication of discrete gradient method for non-smooth optimization
Saeed Banimehri, Hamid Esmaeili
TL;DR
An innovative approach aimed at tackling unconstrained non-smooth optimization problems by introducing a modified discrete gradient method that effectively minimizes the need for evaluations and computational time, ultimately resulting in enhanced efficiency and performance.
- New2026An inexact proximal Newton algorithm with partial Hessian information for nonsmooth separable optimization
Wanyou Cheng, Liang Song, Wenze Deng
From the abstract
We propose a regularized proximal Newton algorithm with partial Hessian information for minimizing the sum of a twice continuously differentiable function and a separable convex (possibly nonsmooth) function.
- New2026Lipschitz-free mirror descent methods for non-smooth optimization problems
Yuan Bowen, Mohammad Soud Alkousa
TL;DR
A Lipschitz-free mirror descent method that achieves weak ergodic convergence is presented, generalizing the convergence results of the mirror descent method in the absence of the Lipschitz assumption.
- New2026Monotone and Nonmonotone Linearized Block Coordinate Descent Methods for Nonsmooth Composite Optimization Problems
Yassine Nabou, Lahcen El Bourkhissi, Sebastian U. Stich et al.
TL;DR
This paper introduces both monotone and nonmonotone variants of LiBCoD, a Linearized Block Coordinate Descent method for solving composite optimization problems, and establishes a global sublinear convergence rate to a stationary point under the assumption of bounded iterates.
- New2026Nonsmooth local optimization via direct search
Christopher John Price, B. L. Robertson
TL;DR
This paper compares the performance of the Sobol, standard Halton and a scrambled Halton poll sequences with no arbitrary step to show the latter performs similarly to the Sobol sequence and outperforms the standard Halton sequence.
- New2026ProxSTORM—a stochastic trust-region algorithm for nonsmooth optimization
Robert Baraldi, Aurya Javeed, Drew Kouri et al.
Read the abstract on the publisher's site - 2025A smooth gradient approximation neural network for general constrained nonsmooth nonconvex optimization problems
Na Liu, Wenwen Jia, Sitian Qin
TL;DR
A smooth gradient approximation neural network is presented for nonconvex optimization problems, in which a smooth approximation technique with time-varying control parameter is introduced for handling nonsmooth nonregular objective functions.
- 2025An improved proximal alternating linearized minimization algorithm and its inertial Bregman extension for structured nonconvex nonsmooth optimization problems on Riemannian manifold
Jiawei Xu, Bo He, Zheng Peng et al.
Overview
Improves the proximal alternating linearized minimization (PALM) algorithm and extends it with inertial and Bregman terms for structured nonconvex, nonsmooth optimization on Riemannian manifolds.
- 2025An inexact ADMM for separable nonconvex and nonsmooth optimization
Jianchao Bai, Miao Zhang, Hongchao Zhang
TL;DR
Numerical experiments on solving nonconvex quadratic programming problems and sparse optimization problems from statistical learning show that the proposed I-ADMM is very effective compared with other state-of-the-art algorithms in the literature.
- 2024A fast smoothing newton method for bilevel hyperparameter optimization for SVC with logistic loss
Yixin Wang, Qingna Li
TL;DR
This paper reformulates the hyperparameter selection for SVC with logistic loss as a bilevel optimization problem in which the upper-level problem and the lower-level problem are both based on logistic loss and shows that the smoothing Newton method has a superlinear convergence rate.
- 2024A modified inertial proximal minimization algorithm for structured nonconvex and nonsmooth problem
Zhonghui Xue, Qianfeng Ma
TL;DR
An enhanced inertial proximal minimization algorithm tailored for a category of structured nonconvex and nonsmooth optimization problems and establishes the global and strong convergence of the algorithmic sequence, contingent upon the assumption that the principal function adheres to the Kurdyka–Łojasiewicz (KL) property.
Multiobjective optimization6
- New2026Optimality conditions in uncertain vector optimization problems with variable domination structures
Yao Lei, QILIN WANG, Madhusudan Das
From the abstract
This paper introduces two new set less order relations based on the Minkowski difference and proposes two novel nonlinear scalarization functionals.
- 2025On quasidifferentiable interval-valued multiobjective optimization
Vivek Laha, Akriti Dwivedi, Prashant Jaiswal
From the abstract
The aim of this article is to investigate approximate solutions in an interval-valued multiobjective optimization problem with inequality constraints involving quasidifferentiable functions, which is denoted by QIVMOP.
- 2025Optimality conditions and duality results for generalized-Hukuhara subdifferentiable preinvex interval-valued vector optimization problems
Zai-Yun Peng, Jianyi Peng, Debdas Ghosh et al.
From the abstract
In this paper, a class of preinvex vector interval optimization problems (VIOP) with gH -subdifferential is considered, and the optimality conditions and dual results are gained.
- 2025Sequential optimality conditions and solution existence for nonsmooth multiobjective optimization problems
Nguyen Le Hoang Anh, Nguyen Canh Hung, Thái Doãn Chương
From the abstract
In this paper, we consider a multiobjective optimization problem, where its feasibility set is unbounded and the related functions are nonsmooth and nonconvex.
- 2024Cardinality-Constrained Multi-objective Optimization: Novel Optimality Conditions and Algorithms
Matteo Lapucci, Pierluigi Mansueto
TL;DR
The concept of L-stationarity is defined and its relationships with other existing conditions and Pareto optimality concepts are analyzed and two novel algorithmic approaches are proposed for solving multi-objective optimization problems with a sparsity constraint on the vector of variables.
- 2024First- and second-order optimality conditions of nonsmooth sparsity multiobjective optimization via variational analysis
Jiawei Chen, Huasheng Su, Xiaoqing Ou et al.
TL;DR
The first-order necessary conditions for local weakly Pareto efficient solution of SMOP are established under some suitable conditions and the equivalence between basic feasible point and stationary point defined by the Fréchet normal cone of SMOP is obtained.
Semidefinite programming5
- 2026A smoothing gradient method for semidefinite programs
Aiqun Huang
Overview
Proposes a smoothing gradient method for solving semidefinite programs.
- 2025A quadratically convergent semismooth Newton method for nonlinear semidefinite programming without generalized Jacobian regularity
Fuxiaoyue Feng, Chao Ding, Xudong Li
From the abstract
We introduce a quadratically convergent semismooth Newton method for nonlinear semidefinite programming that eliminates the need for the generalized Jacobian regularity, a common yet stringent requirement in existing approaches.
- 2025Solving Low-Rank Semidefinite Programs via Manifold Optimization
Jie Wang, Liangbing Hu
TL;DR
A manifold optimization approach to solve linear semidefinite programs (SDP) with low-rank solutions, with an emphasis on SDP relaxations for polynomial optimization problems, and establishes global convergence of the inexact ALM, despite the non-convexity brought by the Burer-Monteiro factorization.
- 2024An augmented Lagrangian-type stochastic approximation method for convex stochastic semidefinite programming defined by expectations
Yule Zhang, Jia Wu, Liwei Zhang
Overview
Develops an augmented Lagrangian-type stochastic approximation method for convex stochastic semidefinite programs whose data are defined by expectations.
- 2024IPRSDP: a primal-dual interior-point relaxation algorithm for semidefinite programming
Ruijin Zhang, Xinwei Liu, Yu-Hong Dai
From the abstract
We propose an efficient primal-dual interior-point relaxation algorithm based on a smoothing barrier augmented Lagrangian, called IPRSDP, for solving semidefinite programming problems in this paper. The IPRSDP algorithm has three advantages over classical interior-point methods.
Convex analysis and optimality conditions5
- 2026Some results of Karush-Kuhn-Tucker multipliers in robust semi-infinite multiobjective optimization
Phan Quốc Khánh, Nguyễn Minh Tùng
From the abstract
We study robust optimization problems which are counterparts of semi-infinite multiobjective nonsmooth and nonconvex optimization problems with data uncertainty.
- 2025Characterization of Generalized FJ and KKT Conditions for Robust Optimization
Javad Koushki, Shokouh Shahbeyk
From the abstract
In this paper, we deal with the generalized Fritz-John (FJ) and Karush-Kuhn-Tucker (KKT) optimality conditions defined by Flores-Bazan and Mastroeni for a nonsmooth nonconvex mathematical programming problem in the face of data uncertainty under the robust counterpart.
- 2025Optimality Conditions at Infinity for Nonsmooth Minimax Programming Problems with Some Applications
Nguyen Van Tuyen, Kwan Deok Bae, Do Sang Kim
From the abstract
This paper is devoted to the study of optimality conditions at infinity in nonsmooth minimax programming problems and their applications.
- 2024Generalized Derivatives and Optimality Conditions in Nonconvex Optimization
Gülçin Dinç Yalçın, Refail Kasımbeyli
From the abstract
In this paper, we study the radial epiderivative notion for nonconvex functions, which extends the (classical) directional derivative concept.
- 2024Necessary and sufficient optimality conditions for nonsmooth nonconvex generalized Nash equilibrium problems
A. Ghasemi, N. Movahedian
From the abstract
The generalized Nash equilibrium problem (GNEP) is an extension of the Nash equilibrium problem (NEP), which allows the strategies and the objective function of each player to depend on the decision variables of all other players.
Machine Learning
51 articlesSupport vector machines and twin SVMs with robust losses, multi-view learning, random vector functional link networks, and the optimization methods used to train machine-learning and deep-learning models.
DeepMathAI work in this area
SVMs, twin SVMs and loss functions5
- 2026Functional data classification using twin support vector machines
Reshma Rastogi, Muskan Singhal
From the abstract
Conventional machine learning algorithms are primarily designed for vector-based data and hence often fail to exploit the inherent continuity and smoothness that characterize functional data.
- 2026Quadratic surface twin support vector machine for imbalanced data
Hossein Moosaei, Milan Hladík, Ahmad Mousavi et al.
TL;DR
A novel approach to tackle binary classification tasks with imbalanced classes by leveraging Universum points to support the minority class within quadratic twin support vector machine models, demonstrating superior performance compared to conventional classifiers and existing methods for imbalanced classification.
- 2025Granular ball twin support vector machine with Universum data
M.A. Ganaie, Vrushank Ahire
TL;DR
The Granular Ball Twin Support Vector Machine with Universum Data (GBU-TSVM) is proposed, which uniquely integrates Universum samples with granular ball computing in the TSVM framework, significantly improving robustness against noise while enhancing computational efficiency.
- 2024Kreĭn twin support vector machines for imbalanced data classification
Cristian Alfonso Jimenez-Castaño, Andrés Marino Álvarez-Meza, David Augusto Cardenas Peña et al.
Overview
Extends twin support vector machines to Kreĭn spaces, which accommodate indefinite kernels, to classify imbalanced data.
- 2024Multi-task twin support vector machine with Universum data
Hossein Moosaei, Fatemeh Bazikar, Milan Hladík
From the abstract
Multi-task learning (MTL) has emerged as a promising topic of machine learning in recent years, aiming to enhance the performance of numerous related learning tasks by exploiting beneficial information.
Multi-view learning11
- New2026A dynamic fusion enhanced trajectory distillation framework for multi-view classification
Hongyu Yao, Binlu Chen, Xiaobin Hong et al.
Read the abstract on the publisher's site - 2026A robust multi-view support vector machine with the RoBoSS loss function
Yash Kumar Arora, Sakshi Gupta, M. Tanveer
TL;DR
A robust multi-view SVM framework that utilizes the RoBoSS loss function called RoBoSS-MvSVM that explicitly integrates consensus and complementarity information across views, enriching the multi-view data representation and ensuring resilient learning is proposed.
- New2026Learning to resolve cross-view conflict for reliable multi-view classification
Yun Liao, Shuyang Hou, Haoyu Wang et al.
Read the abstract on the publisher's site - New2026Multi-type cyberbullying user classification via multi-view feature construction
Tingting Li, Yanlu Gong, Ting Chen et al.
Read the abstract on the publisher's site - New2026Multi-View Contrastive Graph Convolutional Networks with Adaptive Fusion for Semi-Supervised Classification
Huaichun Xiao, Fadi Dornaika, Jingjun Bi et al.
Overview
Proposes multi-view contrastive graph convolutional networks that fuse the views adaptively for semi-supervised classification.
- 2026Multi-view twin random vector functional link for classification problems
Jyoti Maurya, Deepak Gupta
Overview
Introduces a multi-view twin random vector functional link (RVFL) network for classification problems.
- 2026Multiview Ensemble Deep Generalized Eigenvalue Random Vector Functional Link
Yogesh Kumar, Anshika Bhatia, Chavda Jayrajsinh et al.
Overview
Proposes a multi-view ensemble of deep generalized-eigenvalue random vector functional link networks.
- New2026Semi-supervised multi-view classification: A convex formulation and its multi-kernel extension
Jingyu Wang, Yidi Wang, Feiping Nie et al.
Overview
Formulates semi-supervised multi-view classification as a convex problem and extends the formulation to multiple kernels.
- New2026Shapeleter: Shapelet enhancer via multiview positional embedding for time series classification and extrinsic regression
Changchun He, Xin Huo, Baohan Mi et al.
Read the abstract on the publisher's site - 2025Entropy based fuzzy multi-view twin random vector functional link for class imbalanced data
Jyoti Maurya, Deepak Gupta
Overview
Develops an entropy-based fuzzy multi-view twin random vector functional link network for class-imbalanced data.
- 2025Multi-view learning with enhanced multi-weight vector projection support vector machine
Xin Yan, Shuaixing Wang, Huina Chen et al.
TL;DR
Two multi-view enhanced multi-weight vector projection support vector machine models are proposed, each of which tries to generate two projection matrices composed of a set of projection vectors for each view to improve classification performance and efficiency.
Random vector functional link and ensembles5
- 2026Accelerated overrelaxation algorithm for optimizing twin random vector functional link regression
Lang Yu, Nanjing Huang, Wei–Shih Du
From the abstract
This paper proposes a novel twin random vector functional link regression (TRVFLR) model for efficiently addressing regression tasks. TRVFLR extends the classical RVFL framework by incorporating two nonparallel regression functions.
- 2026Affinity-based fuzzy twin random vector functional link network classifier
Chittabarni Sarkar, Deepak Gupta, Rajat Subhra Goswami et al.
Overview
Proposes a fuzzy twin random vector functional link classifier whose fuzzy memberships are based on affinity.
- 2026Deep random vector functional link network for anomaly detection
M. A. Ganaie, Nawab Shahzeb Uddin, Mohammed Wasid et al.
Overview
Applies a deep random vector functional link network to anomaly detection.
- 2024An efficient angle-based twin random vector functional link classifier
Upendra Mishra, Deepak Gupta, Barenya Bikash Hazarika
Overview
Introduces an efficient angle-based twin random vector functional link classifier.
- 2024Domain-incremental learning without forgetting based on random vector functional link networks
ChengRan Liu, Yi Wang, Dong Li et al.
Overview
Uses random vector functional link networks for domain-incremental learning that does not forget earlier domains.
Optimization for machine learning30
Bilevel optimization and hyperparameter selection
- 2026On the Communication Complexity of Decentralized Stochastic Bilevel Optimization
Yihan Zhang, My T. Thai, Jie Wu et al.
TL;DR
This work proposes two novel decentralized stochastic bilevel gradient descent algorithms based on simultaneous and alternating update strategies that can achieve faster convergence rates and lower communication costs than existing methods.
- 2025Decentralized bilevel optimization
Xuxing Chen, Minhui Huang, Shiqian Ma
TL;DR
This paper designs decentralized algorithms for both deterministic and stochastic bilevel optimization problems and analyzes the convergence rates of the proposed algorithms in difference scenarios including the case where data heterogeneity is observed across agents.
- 2025On the Convergence of Proximal Gradient Methods for Convex Simple Bilevel Optimization
Puya Latafat, Andreas Themelis, Silvia Villa et al.
TL;DR
An adaptive linesearch method is developed that introduces a systematic adaptive scheme enabling large and nonmonotonic stepsize sequences while being insensitive to the choice of hyperparameters and initialization.
- 2025Unified smoothing approach for best hyperparameter selection problem using a bilevel optimization strategy
Jan Harold Alcantara, Chieu Thanh Nguyen, Takayuki Okuno et al.
TL;DR
This work proposes new bilevel KKT conditions, which are new necessary optimality conditions tighter than the ones proposed by Okuno et al, and proposes a unified smoothing approach using smoothing functions that belong to the Chen-Mangasarian class.
- 2024A safe screening rule with bi-level optimization of ν support vector machine
Zhiji Yang, Wanyi Chen, Huan Zhang et al.
From the abstract
Support vector machine (SVM) has achieved many successes in machine learning, especially for a small sample problem. As a famous extension of the traditional SVM, the ν support vector machine (ν-SVM) has shown outstanding performance due to its great model interpretability.
- 2024Gradient-based algorithms for multi-objective bi-level optimization
Xinmin Yang, Wei Yao, Haian Yin et al.
TL;DR
This work proposes a gradient-based algorithm for MOBLO, called gMOBA, which has fewer hyperparameters to tune, making it both simple and efficient and demonstrates the theoretical validity by accomplishing the desirable Pareto stationarity.
Deep learning for solving optimization problems (AVE, VI)
- 2026SOR-Unrolled Learning-to-Solve Method for Absolute Value Equations
Cheng-Xi Song, Zi-Yang Zhang, Rui-Zhi Zhou et al.
TL;DR
This paper proposes the first deep unrolled neural network based on SOR-like iteration to solve the AVE, and proposes a specialized network architecture called SOR-Net, which incorporates learnable parameters and nonlinear activation functions into a deep network, with computational layers unrolled from the SOR-like iterations.
- 2024Solving Large-Scale Variational Inequalities with Dynamically Adjusting Initial Condition in Physics-Informed Neural Networks
Dawen Wu, Ludovic Chamoin, Abdel Lisser
Overview
Solves large-scale variational inequalities with physics-informed neural networks whose initial condition is adjusted dynamically.
Dynamical systems and gradient flows of deep-learning training
- 2026Stochastic Dynamical Low-Rank Approximation in the Context of Machine Learning
Arsen Hnatiuk, Jonas Kusch, Lisa Kusch et al.
TL;DR
This work shows that dynamical low-rank training, in combination with stochastic gradient and momentum methods, fulfills certain robustness and descent guarantees and proves convergence to stationary points given the assumption that basis updates are omitted after a certain number of iterations.
- 2025An SDE perspective on stochastic inertial gradient dynamics with time-dependent viscosity and Hessian-driven damping
Rodrigo Maulen-Soto, Jalal M. Fadili, Hédy Attouch et al.
From the abstract
Our approach is part of the close link between continuous dissipative dynamical systems and optimization algorithms. Motivated by solving stochastic convex optimization problems in Hilbert spaces, we propose a class of stochastic inertial differential equations driven by the gradient of the objective function.
- 2024Accelerated optimization in deep learning with a proportional-integral-derivative controller
Song Chen, Jiaxu Liu, Pengkai Wang et al.
TL;DR
A framework for optimization called controlled heavy-ball optimizer, employing the proportional-integral-derivative (PID) controller in the optimizer, and developing a deterministic continuous-time optimizer called Proportional-Integral-Derivative Accelerated Optimizer (PIDAO).
- 2024Efficient and stable SAV-based methods for gradient flows arising from deep learning
Ziqi Ma, Zhiping Mao, Jie Shen
Overview
Develops efficient and stable scalar auxiliary variable (SAV) methods for the gradient flows that arise in training deep-learning models.
Minimax, saddle points and variational inequalities in adversarial learning
- 2026A New Popov’s Method for Solving Quasimonotone Variational Inequalities with Applications in Optimal Control Problems and Machine Learning Algorithms
Olaniyi S. Iyiola, Madushi U. Wickramasinghe, Timilehin Opeyemi Alakoya et al.
From the abstract
This study presents an innovative Popov-type iterative method designed to tackle variational inequality problems (VIPs) with a quasimonotone operator in real Hilbert spaces. Unlike conventional techniques, this method eliminates the need for prior knowledge of the Lipschitz constant of the associated operator.
- 2026An Alternating Gradient Projection Algorithm with Momentum for Nonconvex–Concave Minimax Problems
Jueyou Li, Tao Xie
TL;DR
This paper proposes an alternating gradient projection algorithm with momentum (M-AGP), belonging to single-loop algorithms that not only are easier to implement but also require only the computation of gradient projection updates.
- 2025Calm Local Optimality for Nonconvex-Nonconcave Minimax Problems
Xiaoxiao Ma, Wei Yao, Jane J. Ye et al.
TL;DR
The concept of calm local minimax point is introduced, which is a local minimax point with a calm radius function and it is shown that the calm local minimax optimality and the local minimax optimality coincide under a second-order sufficient condition for the inner maximization problem.
- 2025Single-Loop Projection-Free and Projected Gradient-Based Algorithms for Nonconvex-Concave Saddle Point Problems with Bilevel Structure
Mohammad Mahdi Ahmadi, Erfan Yazdandoost Hamedani
TL;DR
A one-sided projection-free method employing a linear minimization oracle to solve constrained saddle point problems with a bilevel structure, wherein the upper-level objective function is nonconvex-concave and smooth over compact and convex constraint sets, subject to a strongly convex lower-level objective function.
- 2024A Quasi-Newton Subspace Trust Region Algorithm for Nonmonotone Variational Inequalities in Adversarial Learning over Box Constraints
Zicheng Qiu, Jie Jiang, Xiaojun Chen
TL;DR
This paper proposes a quasi-Newton subspace trust region (QNSTR) algorithm for the least squares problems defined by the smoothing approximation of nonsmooth equations and proves the global convergence of the QNSTR algorithm to an ϵ-first-order stationary point of the min-max optimization problem.
Multiobjective optimization in deep learning
- 2026Learning to optimize by multi-gradient for multi-objective optimization
Linxi Yang, Xinmin Yang, Liping Tang
TL;DR
A new automatic learning paradigm for optimizing MOO problems is introduced, and a guarded multi-gradient learning to optimize (GML2O) method is proposed, which automatically learns a generator (or mappings) from multiple gradients to update directions.
- 2025A multiobjective continuation method to compute the regularization path of deep neural networks
Augustina C. Amakor, Konstantin Sonntag, Sebastian Peitz
From the abstract
Sparsity is a highly desired feature in deep neural networks (DNNs) since it ensures numerical efficiency, improves the interpretability (due to the smaller number of relevant features), and robustness.
Nonsmooth and smoothing methods for training neural networks
- 2025Constructing a smoothed Leaky ReLU using a linear combination of the smoothed ReLU and identity function
Meng Si Zhu, Weidong Min, Jiahao Li et al.
TL;DR
Experimental results demonstrate that the proposed general smoothed approximation for the maximum function max(xi,αxi) outperforms the existing state-of-the-art activation functions in improved robustness across different training settings, models, datasets and tasks.
- 2025Convergence analysis of gradient descent based on smoothing L0 regularization and momentum terms
Khidir Shaib Mohamed, Suhail Abdullah Alsaqer, Tahir Bashir et al.
TL;DR
The results of weak convergence for smooth L0 regularization using the adaptive momentum method are presented and it is proven the strong convergence results of the theorems are proven.
- 2025Stochastic Bregman Subgradient Methods for Nonsmooth Nonconvex Optimization Problems
Kuangyu Ding, Kim-Chuan Toh
TL;DR
This paper develops a stochastic Bregman subgradient method that allows the subproblems to be solved inexactly, and introduces a Bregman proximal subgradient method for solving composite optimization problems possibly with constraints, whose convergence can be guaranteed based on the general framework.
- 2024Nonsmooth Nonconvex Stochastic Heavy Ball
Tam Le
TL;DR
This approach builds upon semialgebraic (definable) assumptions commonly met in practical situations and combines a nonsmooth calculus with a differential inclusion method to ensure the convergence of the objective function.
SOCP and SDP for SVMs, feature selection and clustering
- 2026Optimization meets machine learning: an exact algorithm for semi-supervised support vector machines
Veronica Piccialli, Jan Schwiddessen, Antonio M. Sudoso
TL;DR
A new branch-and-cut approach for S3VMs using semidefinite programming (SDP) relaxations is presented, applying optimality-based bound tightening to bound the feasible set and defining a local search heuristic exploiting the solution of the SDP relaxation.
- 2026Semidefinite program-inspired continuous relaxation robust multi-view clustering for large-scale data
Minghao Li, Huiying Xu, Ziying Wang et al.
Overview
Proposes a robust multi-view clustering method for large-scale data, based on a continuous relaxation inspired by semidefinite programming.
- 2026Sparse feature selection via ℓp-quasi-norm second-order cone programming
Miguel Carrasco, Benjamín Ivorra, Julio López et al.
Overview
Performs sparse feature selection with an ℓp quasi-norm model solved through second-order cone programming.
- 2025FSOCP: feature selection via second-order cone programming
Buse Çisil Güldoğuş, Süreyya Akyüz
TL;DR
The proposed FSOCP algorithm trains features individually on a neural network and generates a probability class distribution and prediction, allowing the second-order conic programming model to determine the most important features for improved classification accuracies.
- 2025Pruning the ensemble of convolutional neural networks using second-order cone programming
Buse Çisil Güldoğuş, Abdullah Nazhat Abdullah, Muhammad Ammar Ali et al.
TL;DR
A mathematical model is proposed which prunes the ensemble of Convolutional Neural Networks consisting of different depths and layers that maximizes accuracy and diversity simultaneously with a sparse second order conic optimization model, yielding promising results while reducing model complexity.
- 2024A distributionally robust chance-constrained kernel-free quadratic surface support vector machine
Fengming Lin, Shu‐Cherng Fang, Xiaolei Fang et al.
Overview
Proposes a kernel-free quadratic surface support vector machine that is distributionally robust and chance-constrained, to cope with uncertainty in the data distribution.
- 2024Distributionally robust chance-constrained kernel-based support vector machine
Fengming Lin, Shu‐Cherng Fang, Xiaolei Fang et al.
Overview
Develops a distributionally robust, chance-constrained kernel-based support vector machine for classification under uncertainty in the data distribution.
Generative Models & Inverse Problems
33 articlesGANs and diffusion models for medical imaging, image restoration and deblurring, and stochastic optimization for PET reconstruction.
DeepMathAI work in this area
Generative adversarial networks7
- 2026Purifying medical images with generative adversarial networks
Juexin Zhang, Ying Weng, Boding Wang et al.
From the abstract
Artificial intelligence (AI) applications in medical imaging have the potential to revolutionize healthcare by improving diagnostic accuracy, reducing errors, and increasing efficiency.
- New2026SOM-GAN: A structure-preserving one-to-multiple generative adversarial network for unpaired medical image synthesis
Jinhao Li, Kai Hu, Runze Wang et al.
TL;DR
SOM-GAN is proposed, a Structure-preserving One-to-Multiple Generative Adversarial Network for efficient and structure-preserving image synthesis from a single source domain to multiple target domains and outperforms state-of-the-art methods in both synthesized-image fidelity and downstream clinical segmentation and diagnostic classification tasks.
- New2026Synthesis CT images based on brain MRI images using generative adversarial network with attention mechanisms
Wenxin Li, Jichun Yuan, Bingyang Bao et al.
Overview
Synthesizes CT images from brain MRI with a generative adversarial network that uses attention mechanisms.
- 2025Multimodal medical image fusion combining saliency perception and generative adversarial network
Mohammed Albekairi, Mohamed Vall Ould Mohamed, Khaled Kaâniche et al.
TL;DR
A Temporal Decomposition Network (TDN), a novel deep learning architecture that optimizes multimodal medical image fusion through feature-level temporal analysis and adversarial learning mechanisms, indicates significant potential for clinical applications, particularly in radiological diagnosis, surgical planning, and medical image analysis.
- 2024Generative Adversarial Networks (GANs) for Medical Image Processing: Recent Advancements
Mohammed Mahmood Ali, Mehboob Ali, Mubashir Hussain et al.
TL;DR
This review rigorously evaluates scholarly publications employing Generative Adversarial Networks (GANs) for the synthesis and generation of medical images, segmentation of medical imaging data, image-to-image translation in medical contexts, and denoising or reconstruction of medical imagery.
- 2024ICycle-GAN: Improved cycle generative adversarial networks for liver medical image generation
Ying Chen, Hongping Lin, Wei Zhang et al.
Overview
ICycle-GAN improves cycle-consistent generative adversarial networks for generating liver medical images.
- 2024Survey: application and analysis of generative adversarial networks in medical images
Yang Heng, Ma Yinghua, Fiaz Gul Khan et al.
TL;DR
A comprehensive review of recent research on GANs and their variants in medical applications, including tasks such as image synthesis, segmentation, classification, detection, denoising, reconstruction, fusion, registration, and prediction, is provided.
Diffusion models6
- 2026MAP-based problem-agnostic diffusion model for inverse problems
Pingping Tao, Haixia Liu, Jing Su
From the abstract
Diffusion models have indeed shown great promise in solving inverse problems in image processing. In this paper, we propose a novel, problem-agnostic diffusion model called the maximum a posteriori (MAP)-based guided term estimation method for inverse problems.
- New2026Zero-shot medical image super-resolution using denoising diffusion models with gradient-frequency priors
Jiaqi Zhang, Guo Yang, Rongrong Gao et al.
Overview
Performs zero-shot super-resolution of medical images with denoising diffusion models guided by gradient and frequency priors.
- 2025Deep data consistency: A fast and robust diffusion model-based solver for inverse problems
Hanyu Chen, Zhixiu Hao, Liying Xiao
From the abstract
Diffusion models have become a successful approach for solving various image inverse problems by providing a powerful diffusion prior.
- 2025Fast unconditional diffusion model for accelerated MRI reconstruction
Guijiao Zhao, Chen Zhou, Jianxing Liu et al.
TL;DR
A novel fast diffusion model for MRI reconstruction, termed FDMR, is proposed, which aims to accelerate the inference process and improve reconstruction quality, and operates 4-10 times faster, enabling the reconstruction within just 8 s.
- 2025Unifying and extending diffusion models through PDEs for solving inverse problems
Agnimitra Dasgupta, Alexsander M. Cunha, Ali Fardisi et al.
From the abstract
Diffusion models have emerged as powerful generative tools with applications in computer vision and scientific machine learning (SciML), where they have been used to solve large-scale probabilistic inverse problems.
- 2024DiffGAN: An adversarial diffusion model with local transformer for MRI reconstruction
Xiang Zhao, Tiejun Yang, Bingjie Li et al.
TL;DR
A Local Vision Transformer based adversarial Diffusion model (Diff-GAN) for accelerating MRI reconstruction that employs a generative adversarial network (GAN) as the reverse diffusion model to enable large diffusion steps and mitigates the gradient vanishing issue in GAN training.
Image restoration11
- New2026A paradigm-driven review of diffusion models for image restoration
Zican Lin, Yichi Zhang, Xinyu Yang
Overview
A review of diffusion models for image restoration, organised by the paradigms the methods follow.
- New2026Anti-denoising profile pollution attacks on diffusion model-based recommender systems
Quanqiang Zhou, Tongxue Yang, Keke Wang
Read the abstract on the publisher's site - New2026BD-Diff: Generative Diffusion Model for Image Deblurring on Unknown Domains with Blur-Decoupled Learning
Junhao Cheng, Wei-Ting Chen, Xi Lu et al.
TL;DR
BD-Diff, a generative-diffusion-based model designed to enhance deblurring performance on unknown domains by decoupling structural features and blur patterns through joint training on three specially designed tasks, is proposed.
- New2026DecoupDiff: Spatially decoupled denoising diffusion model for industrial anomaly detection
Liwei Wang, Jinhai Liu, Fengyuan Zuo et al.
Read the abstract on the publisher's site - 2026MSdiff: multi-scale diffusion model for image deblurring
Zhaohan Wang, Chengjun Chen, Chenggang Dai
Overview
MSdiff is a multi-scale diffusion model for image deblurring.
- New2026Perceptual residual denoising diffusion model for image restoration
Jie Yi, Yunyun Yang, Lin Zhang
Read the abstract on the publisher's site - New2026Retraining-free multi-task MR image restoration using a pretrained diffusion model with Bayesian optimization
N. Fujita, Yasuhiko Terada
TL;DR
The results suggest the feasibility of adapting a single pretrained diffusion prior across multiple MR image restoration tasks through guidance calibration without task-specific weight retraining, while indicating that robustness to adaptation-case selection and computational cost require further investigation.
- 2025CDRM: Controllable diffusion restoration model for realistic image deblurring
Ziyi Chen, Guangmang Cui, Jufeng Zhao et al.
Overview
CDRM is a controllable diffusion restoration model for deblurring real-world images.
- 2025Diffusion Models for Image Restoration and Enhancement: A Comprehensive Survey
Xin Li, Yulin Ren, Xin Jin et al.
TL;DR
This first to present a comprehensive review of recent diffusion model-based methods on image restoration, encompassing the learning paradigm, conditional strategy, framework design, modeling strategy, and evaluation, and proposes nine potential and challenging directions for the future research of diffusion model-based IR.
- 2024A new diffusion method for blind image denoising
Yonggui Zhu, Yaling Chen
TL;DR
A novel perspective for blind denoising is proposed by converting the static image denoising problem into a dynamic process inspired by the diffusion model, which outperforms existing methods and achieves state-of-the-art results in blind additive white Gaussian noise and real-world image denoising.
- 2024Hierarchical wavelet-guided diffusion model for single image deblurring
Xiaopan Li, Shiqian Wu, Xin Yuan et al.
From the abstract
Image deblurring is a critical preprocessing step in various high-level vision tasks, including facial recognition, medical imaging, and object detection. Blurred artifacts can arise from multiple factors, such as camera shakes and fast-moving objects.
PET reconstruction and stochastic optimization9
- 2026A preconditioned proximal gradient algorithm for PET image reconstruction with approximately sparse regularization
Yongxin He, Yizun Lin, Manchun Tan et al.
Overview
Proposes a preconditioned proximal gradient algorithm for PET image reconstruction with an approximately sparse regularization term.
- New2026Deep learning enables quantitative kinetic modeling from low-dose dynamic [$$^{18}$$F]-MK6240 Tau PET
Se‐In Jang, Yassir Najmaoui, Yanis Chemli et al.
TL;DR
The results support that a 90% dose reduction is possible without losing the quantitative accuracy needed for tau burden assessment in clinical and research settings.
- New2026Deep learning-based enhancement of ultra-low-dose multi-tracer PET/MR: a technical feasibility study in Alzheimer’s disease
Na Qi, Boyang Pan, Zengbei Yuan et al.
TL;DR
The DDWAN model improved image quality and quantitative consistency of simulated ultra-low-dose multi-tracer PET/MR images compared with full-dose references acquired under the same protocol, which support technical feasibility in Alzheimer’s disease (AD).
- New2026Quantitative accuracy and clinical utility of ultra-low dose CT protocols for attenuation correction in PET/CT
Georgios I. Angelis, Yaser Hadi Gholami, Kathy P. Willowson et al.
TL;DR
Tin-filtered ultra-low-dose CT enables genuinely low-dose whole-body PET/CT, achieving total effective doses below 1 mSv while preserving quantitative accuracy, lesion conspicuity, and clinical interpretability.
- New2026Toward an optimized PET reconstruction protocol for spinal cord: superiority of OSEM with TOF and PSF
Jiyuan Wang, Jing Huang, Bixiao Cui et al.
TL;DR
The OSEM + PSF + TOF reconstruction method demonstrated significant advantages in both image quality and quantitative assessment of FDG uptake, making it the recommended choice for spinal cord PET image reconstruction.
- 2025Advanced deep learning for PET image reconstruction and quality improvement
Asma Benyelles, Amel Korti
TL;DR
This study aims to improve image quality and eliminate residual noise from PET images by applying an efficient Generative Adversarial Network (GAN) model to sinograms before reconstruction using the Ordered Subsets Expectation Maximization (OSEM), an effective method for high-quality data reconstruction.
- 2025FastDIP: An effective approach for accelerating unsupervised low-count PET image reconstruction
Jinming Li, Jing Wang, Yang Lv et al.
TL;DR
FastDIP is an efficient approach for unsupervised low-count PET image reconstruction that significantly reduces the network training time and markedly enhances image restoration performance.
- 2024Stochastic model for enhanced PET image reconstruction
Tomislav Matulić, Damir Seršić
TL;DR
It is proved, experimentally and analytically, that the difference between the best approximations and real crystal-to-crystal response is insignificant.
- 2024Trapezoidal back projection for positron emission tomography reconstruction
Dóra Varnyú, Krisztián Paczári, László Szirmay‐Kalos
TL;DR
The proposed deterministic TBP algorithm achieves a low noise level in a short runtime, thus it can be a promising solution for the back projection of the 3D PET reconstruction.
Medical AI
12 articlesDeep learning and radiomics for vertebral fractures and osteoporosis screening from CT, MRI and radiographs.
DeepMathAI work in this area
Spine and vertebral fractures5
- 2026Deep learning magnetic resonance imaging algorithm for differentiating metastatic vertebral fractures
Joonghyun Ahn, Young-Hoon Kim, Sang-Il Kim et al.
TL;DR
A deep learning-based object detection approach can accurately distinguish malignant metastatic lesions from benign VCFs on routine MRI and may serve as a practical decision-support tool to expedite oncologic referral, improve diagnostic accuracy, and optimize treatment strategies.
- 2026Deep Learning-Assisted Radiomics for Predicting Thoracolumbar Osteoporotic Vertebral Fractures
Maggie Che, Y Q Chen, Mohamad Noor Shafini et al.
TL;DR
The findings demonstrate that radiomics derived from routine non-contrast CT, leveraging deep learning for automated bone compartment segmentation, offers a superior and practical tool for Early identification of high-risk individuals for thoracolumbar OVFs.
- 2025A deep learning pipeline for systematic and accurate vertebral fracture reporting in computed tomography
Carl Glessgen, Joshy Cyriac, S. Yang et al.
TL;DR
A deep learning pipeline was built and shown to accurately detect fractures on CT images of lumbar/thoracolumbar spine images of varying field-of-views.
- 2024A Deep-Learning Model for Diagnosing Fresh Vertebral Fractures on Magnetic Resonance Images
Yanni Wang, Gang Liu, Lei Wang et al.
TL;DR
The findings highlighted the potential of AI in diagnosing fresh VFs, offering an accurate and efficient way to aid physicians with diagnosis and treatment decisions.
- 2024Deep learning model for automated detection of fresh and old vertebral fractures on thoracolumbar CT
Jianan Chen, Song Liu, Yong Li et al.
TL;DR
The deep learning model can automatically and accurately segment the vertebral bodies with compression fractures and classify them as fresh or old fractures, thereby assisting clinicians in making clinical decisions.
Osteoporosis7
- 2026Comparison of dual-energy X-ray absorptiometry and lumbar radiographs using deep learning for osteoporosis diagnosis
Deniz Apalan, Bülent Yildiz, Özlem Polat et al.
TL;DR
DL analysis of lumbar radiographs, combined with transfer learning and metaheuristic optimization, may support BMD category classification as a complementary screening tool for osteoporosis screening and diagnosis.
- New2026Methodological considerations on opportunistic screening for osteoporosis using chest X-rays and deep learning
Gde Bagus Susila Pradnyana Nurcahyo Oka
Overview
Discusses methodological considerations in opportunistic osteoporosis screening that applies deep learning to chest X-rays.
- New2026Optimizing deep learning-enhanced opportunistic osteoporosis screening from chest X-rays in Australia: Impact of follow-up and treatment adherence on clinical and economic outcomes
Jean‐Yves Reginster, Peter Robert Ebeling, Majed Alokail et al.
TL;DR
It is highlighted that follow-up and persistence are not merely complementary but essential for realizing the full potential of cost-effective AI-driven osteoporosis screening, and Optimizing follow-up and adherence could amplify clinical benefits by four- to five-fold.
- 2024A new superfluity deep learning model for detecting knee osteoporosis and osteopenia in X-ray images
Soaad M. Naguib, Mohammed Khalid Saleh, Hanaa M. Hamza et al.
TL;DR
The results indicate that the performance of the pre-trained models, namely AlexNet and ResNet50, was inferior to that of the proposed Superfluity DL architecture, and the Superfluity DL model demonstrated the highest accuracy among all the pre-trained models.
- 2024Application of Machine Learning to Osteoporosis and Osteopenia Screening Using Hand Radiographs
Anna Luan, Zeshaan N. Maan, Kun-Yi Lin et al.
TL;DR
A deep learning algorithm was developed using the ResNet-50 framework and trained to predict the presence of osteoporosis or osteopenia on hand radiographs using labeled images, representing a possible step toward more accessible, cost-effective, automated diagnosis and therefore earlier treatment of osteoporosis/osteopenia.
- 2024Deep learning for osteoporosis screening using an anteroposterior hip radiograph image
Artit Boonrod, Prarinthorn Piyaprapaphan, Nut Kittipongphat et al.
TL;DR
The results suggest that the deep learning model might potentially be used as a screening tool to find patients at risk for osteoporosis to perform further BMD measurement, and the best model was determined by the area under the curves (AUC).
- 2024Predicting osteoporosis from kidney-ureter-bladder radiographs utilizing deep convolutional neural networks
Tzu-Yun Yen, Chan-Shien Ho, Yu‐Cheng Pei et al.
TL;DR
The proposed DeepDXA-KUB, a deep learning model that predicts the BMD values of the left hip and lumbar vertebrae from an input KUB image, is the first to develop a deep learning model based on KUB radiographs to predict lumbar spine and femoral BMD.
Curated by the DeepMathAI Research Group from Scopus Q1–Q2 journals (quartiles from SJR); every week, recent articles on each topic from the same journals are added automatically and marked New. Bibliographic data and open-access links from OpenAlex. Summaries are an AI-generated TL;DR from Semantic Scholar, the opening sentences of the article's abstract, or a short overview written for this site from the article's title where no abstract is freely available. Links lead to the publishers and to open repositories such as arXiv; this site does not host the articles, and some publisher pages require a subscription.