arXiv subjects

Heng Zhu

Publications and source records attributed to Heng Zhu.

14 recordsLinked to original sources

arXiv2026

Storage-Scalable Progressive Semantic Communication via Knowledge-Base Reuse

Existing knowledge-base-assisted semantic communication schemes commonly adopt either single knowledge-base quantization (SKBQ) or multi-knowledge-base residual quantization (MKBQ). SKBQ incurs limited storage overhead but has restricted quantization capacity, whereas MKBQ supports progressive refinement by assigning an independent knowledge base (KB) to each stage, causing the KB storage to grow linearly with the transmission depth. To address this problem, we propose storage-scalable knowledge-base reuse quantization (SSKBQ), which reuses a compact set of KBs across multiple residual refinement stages and thereby decouples the number of transmission stages from the number of maintained KBs. A stage-aware residual supervision mechanism is further introduced to regularize intermediate quantized representations and encourage progressive refinement. Experimental results demonstrate that KB reuse provides an effective solution to the storage scalability problem while maintaining competitive progressive reconstruction performance.

cs.LG↗

arXiv2026

Understanding From Human Perspective: A Multi-agent System for Interactive Egocentric Medical Image Segmentation

Interactive egocentric medical image segmentation (IEMIS) plays an important role in smart-glasses-assisted medical image review, segmenting the medical targets a clinician refers to from their egocentric view. Once it succeeds, the object-level visual evidence it provides strengthens the review and underpins fine-grained analysis and clinical decision-making. However, the instruction and the video both come from the user's egocentric perspective, which poses two challenges. (1) Semantic ambiguity leaves the model unable to confirm the user-intended target. (2) Visual variability makes the segmentation jump from frame to frame. In this paper, we propose EgoMed-Agent, a multi-agent system that understands the target from the human perspective through two workflows. (1) The \textit{Target Confirmation Workflow} grounds the instruction against candidate targets with a reliability score, confirming the target when the grounding is reliable and asking the user to clarify when it is not, thereby confirming the segmentation target. (2) The \textit{Localization-Guided Propagation Workflow} couples mask propagation with per-frame target localization, using the localized target to correct the propagated mask whenever the two diverge, so the segmentation stays on the target across the egocentric video. Extensive experiments show that EgoMed-Agent reaches 71.34\% average Dice, far above the best text-prompted baseline (11.70\%). Our code is available at \href{https://github.com/wdyyyyyy/EgoMed-Agent}{our project page}.

cs.CV↗

arXiv2025

LocalKMeans: Convergence of Lloyd's Algorithm with Distributed Local Iterations

In this paper, we analyze the classical $K$-means alternating-minimization algorithm, also known as Lloyd's algorithm (Lloyd, 1956), for a mixture of Gaussians in a data-distributed setting that incorporates local iteration steps. Assuming unlabeled data distributed across multiple machines, we propose an algorithm, LocalKMeans, that performs Lloyd's algorithm in parallel in the machines by running its iterations on local data, synchronizing only every $L$ of such local steps. We characterize the cost of these local iterations against the non-distributed setting, and show that the price paid for the local steps is a higher required signal-to-noise ratio. While local iterations were theoretically studied in the past for gradient-based learning methods, the analysis of unsupervised learning methods is more involved owing to the presence of latent variables, e.g. cluster identities, than that of an iterative gradient-based algorithm. To obtain our results, we adapt a virtual iterate method to work with a non-convex, non-smooth objective function, in conjunction with a tight statistical analysis of Lloyd steps.

stat.ML↗

arXiv2024

Effectiveness of Distributed Gradient Descent with Local Steps for Overparameterized Models

In distributed training of machine learning models, gradient descent with local iterative steps, commonly known as Local (Stochastic) Gradient Descent (Local-(S)GD) or Federated averaging (FedAvg), is a very popular method to mitigate communication burden. In this method, gradient steps based on local datasets are taken independently in distributed compute nodes to update the local models, which are then aggregated intermittently. In the interpolation regime, Local-GD can converge to zero training loss. However, with many potential solutions corresponding to zero training loss, it is not known which solution Local-GD converges to. In this work we answer this question by analyzing implicit bias of Local-GD for classification tasks with linearly separable data. For the interpolation regime, our analysis shows that the aggregated global model obtained from Local-GD, with arbitrary number of local steps, converges exactly to the model that would be obtained if all data were in one place (centralized model) ''in direction''. Our result gives the exact rate of convergence to the centralized model with respect to the number of local steps. We also obtain the same implicit bias with a learning rate independent of number of local steps with a modified version of the Local-GD algorithm. Our analysis provides a new view to understand why Local-GD can still perform well with a very large number of local steps even for heterogeneous data. Lastly, we also discuss the extension of our results to Local-SGD and non-separable data.

cs.LG↗

arXiv2024

Photoinduced Topological Phase Transitions in a Kitaev kagome magnet

The photoinduced topological phase transitions and thermal Hall conductivity of a kagome magnet with Heisenberg, Kitaev, and DM interactions under an external magnetic field aligned along the [111] directions is investigated in this study. In the presence of a strong magnetic field perpendicular to the lattice plane in the [111] direction, the system exhibits a fully polarized paramagnetic phase and the magnon band carries an asymmetric Chern number across the phase diagram region. Utilizing magnetic Floquet-Bloch theory, we demonstrate that periodically driven intrinsic topological magnetic materials can be manipulated into different topological phases with varying Berry curvature, Chern numbers and thermal Hall conductivities signatures by adjusting light intensity throughout the phase diagram region.

cond-mat.str-el↗

arXiv2024

Optimal Compression of Unit Norm Vectors in the High Distortion Regime

Motivated by the need for communication-efficient distributed learning, we investigate the method for compressing a unit norm vector into the minimum number of bits, while still allowing for some acceptable level of distortion in recovery. This problem has been explored in the rate-distortion/covering code literature, but our focus is exclusively on the "high-distortion" regime. We approach this problem in a worst-case scenario, without any prior information on the vector, but allowing for the use of randomized compression maps. Our study considers both biased and unbiased compression methods and determines the optimal compression rates. It turns out that simple compression schemes are nearly optimal in this scenario. While the results are a mix of new and known, they are compiled in this paper for completeness.

cs.IT↗

arXiv2023

Interacting Floquet topological magnons in laser-irradiated Heisenberg honeycomb ferromagnets

When a Heisenberg honeycomb ferromagnet is irradiated by high frequency circularly polarized light, the underlying uncharged magnons acquire a time dependent Aharonov Casher phase, which makes it a Floquet topological magnon insulator. In this context, we investigate the many body interaction effects of Floquet magnons in laser irradiated Heisenberg honeycomb ferromagnets with ocontaining Dzyaloshinskii Moriya interaction under the application of circularly polarized off resonant light. We demonstrate that the quantum ferromagnet systems periodically laser driven exhibits temperature driven topological phase transitions due to Floquet magnon magnon interactions. The thermal Hall effect of Floquet magnons serves as a prominent signature for detecting these many body effects near the critical point, enabling experimental investigation into this phenomenon. Our study complements the lack of previous theoretical works that the topological phase transition of the Floquet magnon under the linear spin wave approximation is only tunable by the light field. Our study presents a novel approach for constructing Floquet topological phases in periodically driven quantum magnet systems that goes beyond the limitations of the linear spin wave theory. We provide numerical results based on the well known van der Waals quantum magnet CrX3 (X=F, Cl, Br, and I), calling for experimental implementation.

cond-mat.mes-hall↗

arXiv2023

On the Tradeoff between Privacy Preservation and Byzantine-Robustness in Decentralized Learning

This paper jointly considers privacy preservation and Byzantine-robustness in decentralized learning. In a decentralized network, honest-but-curious agents faithfully follow the prescribed algorithm, but expect to infer their neighbors' private data from messages received during the learning process, while dishonest-and-Byzantine agents disobey the prescribed algorithm, and deliberately disseminate wrong messages to their neighbors so as to bias the learning process. For this novel setting, we investigate a generic privacy-preserving and Byzantine-robust decentralized stochastic gradient descent (SGD) framework, in which Gaussian noise is injected to preserve privacy and robust aggregation rules are adopted to counteract Byzantine attacks. We analyze its learning error and privacy guarantee, discovering an essential tradeoff between privacy preservation and Byzantine-robustness in decentralized learning -- the learning error caused by defending against Byzantine attacks is exacerbated by the Gaussian noise added to preserve privacy. For a class of state-of-the-art robust aggregation rules, we give unified analysis of the "mixing abilities". Building upon this analysis, we reveal how the "mixing abilities" affect the tradeoff between privacy preservation and Byzantine-robustness. The theoretical results provide guidelines for achieving a favorable tradeoff with proper design of robust aggregation rules. Numerical experiments are conducted and corroborate our theoretical findings.

cs.LG↗

arXiv2023

Photoinduced topological phase transitions in Kitaev-Heisenberg honeycomb ferromagnets with the Dzyaloshinskii-Moriya interaction

We theoretically study topological properties of Floquet magnon in a laser-irradiated Kitaev-Heisenberg honeycomb ferromagnet with the Dzyaloshinskii-Moriya interaction by means of the Floquet-Bloch theory. It is found that the Kitaev-Heisenberg ferromagnet can reveal two topological phases with different Chern numbers when it is irradiated by a circular-polarized light laser. Our results show that the topological phase of the system can be switched from one topological phase to another one via varying the light intensity. The intrinsic DMI plays a crucial role in the occurrence of photoinduced topological phase transition. It is shown that the sign reversal of the thermal hall conductivity is an important indicator on photoinduced topological phase transitions in the Kitaev-Heisenberg honeycomb ferromagnet.

cond-mat.str-el↗

arXiv2023

Interacting topological magnons in a checkerboard ferromagnet

This work is devoted to studying the magnon-magnon interaction effect in a two-dimensional checkerboard ferromagnet with the Dzyaloshinskii-Moriya interaction. By means of the first-order Green function formalism, the influence of magnon-magnon interaction on the magnon band topology is analyzed. In order to verify that the gap-closing phenomenon is a signature for the topological phase transitions of the checkerboard ferromagnet, we display that the Chern numbers of renormalized magnon bands are distinct above and below the critical temperature. Our results show that the checkerboard ferromagnet possesses two topological phases and its topological phase can be controlled either by the temperature or applied magnetic field due to magnon-magnon interactions. Interestingly, we find that the topological phase transition occurs twice with the increase of the temperature, which is different from the result of the honeycomb ferromagnet.

cond-mat.str-el↗

arXiv2023

Topological phase transitions in a honeycomb ferromagnet with unequal Dzyaloshinskii-Moriya interactions

This theoretical research is devoted to study topological phase transitions in a two-dimensional honeycomb ferromagnetic lattice with unequal Dzyaloshinskii-Moriya interactions for the two sublattices. With the help of a first-order Green function formalism, we analyze the influence of magnon-magnon interaction on the magnon band topology. It is found that the existence of the antichiral Dzyaloshinskii-Moriya interaction can led to a tilting of the renormalized magnon bands near the Dirac momenta. Then, the renormalized magnon band gaps at Dirac points have different widths. Through changing the temperature, we can observe the renormalized magnon band gap closing-reopening phenomenon, which corresponds to the topological phase transition. Our results show that the critical temperature of the topological phase transition is related to the strength of the antichiral Dzyaloshinskii-Moriya interaction.

cond-mat.other↗

arXiv2022

Intrinsic localized modes in a two-dimensional checkerboard ferromagnetic lattice

An analytical work on intrinsic localized modes in a two-dimensional Heisenberg ferromagnet on the checkerboard lattice is presented. Taking advantage of an asymptotic method, the governing lattice dynamical equations are reduced to one (2+1) -dimensional nonlinear Schrödinger. In our work, we obtain two types of nonlinear localized mode solutions, namely, Brillouin zone center modes and Brillouin zone corner modes. The occurrence conditions for these intrinsic localized modes are given in detail. Especially, we find that the competition between the Dzialozinskii-Moriy interaction and the next-nearest neighbor interaction of the checkerboard ferromagnet has an effect on the local structure of the Brillouin zone corner acoustic mode.

nlin.PS↗

arXiv2022

Bridging Differential Privacy and Byzantine-Robustness via Model Aggregation

This paper aims at jointly addressing two seemly conflicting issues in federated learning: differential privacy (DP) and Byzantine-robustness, which are particularly challenging when the distributed data are non-i.i.d. (independent and identically distributed). The standard DP mechanisms add noise to the transmitted messages, and entangles with robust stochastic gradient aggregation to defend against Byzantine attacks. In this paper, we decouple the two issues via robust stochastic model aggregation, in the sense that our proposed DP mechanisms and the defense against Byzantine attacks have separated influence on the learning performance. Leveraging robust stochastic model aggregation, at each iteration, each worker calculates the difference between the local model and the global one, followed by sending the element-wise signs to the master node, which enables robustness to Byzantine attacks. Further, we design two DP mechanisms to perturb the uploaded signs for the purpose of privacy preservation, and prove that they are $(ε,0)$-DP by exploiting the properties of noise distributions. With the tools of Moreau envelop and proximal point projection, we establish the convergence of the proposed algorithm when the cost function is nonconvex. We analyze the trade-off between privacy preservation and learning performance, and show that the influence of our proposed DP mechanisms is decoupled with that of robust stochastic model aggregation. Numerical experiments demonstrate the effectiveness of the proposed algorithm.

cs.LG↗

arXiv2022

BROADCAST: Reducing Both Stochastic and Compression Noise to Robustify Communication-Efficient Federated Learning

Communication between workers and the master node to collect local stochastic gradients is a key bottleneck in a large-scale federated learning system. Various recent works have proposed to compress the local stochastic gradients to mitigate the communication overhead. However, robustness to malicious attacks is rarely considered in such a setting. In this work, we investigate the problem of Byzantine-robust compressed federated learning, where the attacks from Byzantine workers can be arbitrarily malicious. We theoretically point out that different to the attacks-free compressed stochastic gradient descent (SGD), its vanilla combination with geometric median-based robust aggregation seriously suffers from the compression noise in the presence of Byzantine attacks. In light of this observation, we propose to reduce the compression noise with gradient difference compression so as to improve the Byzantine-robustness. We also observe the impact of the intrinsic stochastic noise caused by selecting random samples, and adopt the stochastic average gradient algorithm (SAGA) to gradually eliminate the inner variations of regular workers. We theoretically prove that the proposed algorithm reaches a neighborhood of the optimal solution at a linear convergence rate, and the asymptotic learning error is in the same order as that of the state-of-the-art uncompressed method. Finally, numerical experiments demonstrate the effectiveness of the proposed method.

cs.LG↗