arXiv subjects

Kexin Jin

Publications and source records attributed to Kexin Jin.

12 recordsLinked to original sources

arXiv2025

Tunable Two-Dimensional Electron Gas at the Interfaces of Ferroelectric Potassium Tantalate Niobates

The heterointerfaces at complex oxides have emerged as a promising platform for discovering novel physical phenomena and advancing integrated sensing, storage, and computing technologies. Nevertheless, achieving precise control over a two-dimensional electron gas (2DEG) in a ferroelectric oxide-based field-effect transistor (FET) configuration remains challenging. Here, we firstly demonstrate a tunable 2DEG system fabricated by depositing an amorphous LaAlO3(LAO) film onto a (001)-oriented ferroelectric potassium tantalate niobate substrate. Interfaces grown under high-temperature and high-oxygen-pressure conditions exhibit a good metallic conduction. Notably, well-defined metallic 2DEGs displaying pronounced hysteresis and persistent electric-field-modulated resistance are observed below 108 K, achieving a resistance modulation of 11.6% at 7 K. These results underscore the potential for extending such behavior to other oxide-based 2DEG systems and facilitate further exploration of ferroelectric metals in complex oxide heterostructures.

cond-mat.mtrl-sci↗

arXiv2024

Losing momentum in continuous-time stochastic optimisation

The training of modern machine learning models often consists in solving high-dimensional non-convex optimisation problems that are subject to large-scale data. In this context, momentum-based stochastic optimisation algorithms have become particularly widespread. The stochasticity arises from data subsampling which reduces computational cost. Both, momentum and stochasticity help the algorithm to converge globally. In this work, we propose and analyse a continuous-time model for stochastic gradient descent with momentum. This model is a piecewise-deterministic Markov process that represents the optimiser by an underdamped dynamical system and the data subsampling through a stochastic switching. We investigate longtime limits, the subsampling-to-no-subsampling limit, and the momentum-to-no-momentum limit. We are particularly interested in the case of reducing the momentum over time. Under convexity assumptions, we show convergence of our dynamical system to the global minimiser when reducing momentum over time and letting the subsampling rate go to infinity. We then propose a stable, symplectic discretisation scheme to construct an algorithm from our continuous-time dynamical system. In experiments, we study our scheme in convex and non-convex test problems. Additionally, we train a convolutional neural network in an image classification problem. Our algorithm {attains} competitive results compared to stochastic gradient descent with momentum.

math.OC↗

arXiv2024

How to beat a Bayesian adversary

Deep neural networks and other modern machine learning models are often susceptible to adversarial attacks. Indeed, an adversary may often be able to change a model's prediction through a small, directed perturbation of the model's input - an issue in safety-critical applications. Adversarially robust machine learning is usually based on a minmax optimisation problem that minimises the machine learning loss under maximisation-based adversarial attacks. In this work, we study adversaries that determine their attack using a Bayesian statistical approach rather than maximisation. The resulting Bayesian adversarial robustness problem is a relaxation of the usual minmax problem. To solve this problem, we propose Abram - a continuous-time particle system that shall approximate the gradient flow corresponding to the underlying learning problem. We show that Abram approximates a McKean-Vlasov process and justify the use of Abram by giving assumptions under which the McKean-Vlasov process finds the minimiser of the Bayesian adversarial robustness problem. We discuss two ways to discretise Abram and show its suitability in benchmark adversarial deep learning experiments.

cs.LG↗

arXiv2024

Subsampling Error in Stochastic Gradient Langevin Diffusions

The Stochastic Gradient Langevin Dynamics (SGLD) are popularly used to approximate Bayesian posterior distributions in statistical learning procedures with large-scale data. As opposed to many usual Markov chain Monte Carlo (MCMC) algorithms, SGLD is not stationary with respect to the posterior distribution; two sources of error appear: The first error is introduced by an Euler--Maruyama discretisation of a Langevin diffusion process, the second error comes from the data subsampling that enables its use in large-scale data settings. In this work, we consider an idealised version of SGLD to analyse the method's pure subsampling error that we then see as a best-case error for diffusion-based subsampling MCMC methods. Indeed, we introduce and study the Stochastic Gradient Langevin Diffusion (SGLDiff), a continuous-time Markov process that follows the Langevin diffusion corresponding to a data subset and switches this data subset after exponential waiting times. There, we show the exponential ergodicity of SLGDiff and that the Wasserstein distance between the posterior and the limiting distribution of SGLDiff is bounded above by a fractional power of the mean waiting time. We bring our results into context with other analyses of SGLD.

stat.ML↗

arXiv2023

DSGD-CECA: Decentralized SGD with Communication-Optimal Exact Consensus Algorithm

Decentralized Stochastic Gradient Descent (SGD) is an emerging neural network training approach that enables multiple agents to train a model collaboratively and simultaneously. Rather than using a central parameter server to collect gradients from all the agents, each agent keeps a copy of the model parameters and communicates with a small number of other agents to exchange model updates. Their communication, governed by the communication topology and gossip weight matrices, facilitates the exchange of model updates. The state-of-the-art approach uses the dynamic one-peer exponential-2 topology, achieving faster training times and improved scalability than the ring, grid, torus, and hypercube topologies. However, this approach requires a power-of-2 number of agents, which is impractical at scale. In this paper, we remove this restriction and propose \underline{D}ecentralized \underline{SGD} with \underline{C}ommunication-optimal \underline{E}xact \underline{C}onsensus \underline{A}lgorithm (DSGD-CECA), which works for any number of agents while still achieving state-of-the-art properties. In particular, DSGD-CECA incurs a unit per-iteration communication overhead and an $\tilde{O}(n^3)$ transient iteration complexity. Our proof is based on newly discovered properties of gossip weight matrices and a novel approach to combine them with DSGD's convergence analysis. Numerical experiments show the efficiency of DSGD-CECA.

cs.LG↗

arXiv2023

AdaNPC: Exploring Non-Parametric Classifier for Test-Time Adaptation

Many recent machine learning tasks focus to develop models that can generalize to unseen distributions. Domain generalization (DG) has become one of the key topics in various fields. Several literatures show that DG can be arbitrarily hard without exploiting target domain information. To address this issue, test-time adaptive (TTA) methods are proposed. Existing TTA methods require offline target data or extra sophisticated optimization procedures during the inference stage. In this work, we adopt Non-Parametric Classifier to perform the test-time Adaptation (AdaNPC). In particular, we construct a memory that contains the feature and label pairs from training domains. During inference, given a test instance, AdaNPC first recalls K closed samples from the memory to vote for the prediction, and then the test feature and predicted label are added to the memory. In this way, the sample distribution in the memory can be gradually changed from the training distribution towards the test distribution with very little extra computation cost. We theoretically justify the rationality behind the proposed method. Besides, we test our model on extensive numerical experiments. AdaNPC significantly outperforms competitive baselines on various DG benchmarks. In particular, when the adaptation target is a series of domains, the adaptation accuracy of AdaNPC is 50% higher than advanced TTA methods. The code is available at https://github.com/yfzhang114/AdaNPC.

cs.LG↗

arXiv2023

Communication-Efficient Topologies for Decentralized Learning with $O(1)$ Consensus Rate

Decentralized optimization is an emerging paradigm in distributed learning in which agents achieve network-wide solutions by peer-to-peer communication without the central server. Since communication tends to be slower than computation, when each agent communicates with only a few neighboring agents per iteration, they can complete iterations faster than with more agents or a central server. However, the total number of iterations to reach a network-wide solution is affected by the speed at which the agents' information is ``mixed'' by communication. We found that popular communication topologies either have large maximum degrees (such as stars and complete graphs) or are ineffective at mixing information (such as rings and grids). To address this problem, we propose a new family of topologies, EquiTopo, which has an (almost) constant degree and a network-size-independent consensus rate that is used to measure the mixing efficiency. In the proposed family, EquiStatic has a degree of $Θ(\ln(n))$, where $n$ is the network size, and a series of time-dependent one-peer topologies, EquiDyn, has a constant degree of 1. We generate EquiDyn through a certain random sampling procedure. Both of them achieve an $n$-independent consensus rate. We apply them to decentralized SGD and decentralized gradient tracking and obtain faster communication and better convergence, theoretically and empirically. Our code is implemented through BlueFog and available at \url{https://github.com/kexinjinnn/EquiTopo}

math.OC↗

arXiv2021

A Continuous-time Stochastic Gradient Descent Method for Continuous Data

Optimization problems with continuous data appear in, e.g., robust machine learning, functional data analysis, and variational inference. Here, the target function is given as an integral over a family of (continuously) indexed target functions - integrated with respect to a probability measure. Such problems can often be solved by stochastic optimization methods: performing optimization steps with respect to the indexed target function with randomly switched indices. In this work, we study a continuous-time variant of the stochastic gradient descent algorithm for optimization problems with continuous data. This so-called stochastic gradient process consists in a gradient flow minimizing an indexed target function that is coupled with a continuous-time index process determining the index. Index processes are, e.g., reflected diffusions, pure jump processes, or other Lévy processes on compact spaces. Thus, we study multiple sampling patterns for the continuous data space and allow for data simulated or streamed at runtime of the algorithm. We analyze the approximation properties of the stochastic gradient process and study its longtime behavior and ergodicity under constant and decreasing learning rates. We end with illustrating the applicability of the stochastic gradient process in a polynomial regression problem with noisy functional data, as well as in a physics-informed neural network.

cs.LG↗

arXiv2021

Large positive magnetoresistance in photocarrier doping potassium tantalites in the extreme quantum limit

We report on a high-field magnetotransport study of KTaO3 single crystals. This material is a promising candidate to study in the extreme quantum limit (EQL). By photocarrier doping with 360 nm light, we have observed a significant positive, non-saturating, and linear magnetoresistance at low temperatures accompanied by a vanishing Hall coefficient. When cooled down to 10 K and subjected to a magnetic field of 12 T, the value of magnetoresistance of KTaO3 (100) is increased by as much as 433%. Such behavior can be attributed to all electrons occupying only the lowest Landau level in the EQL. In this state, a quantum magnetoresistance is produced. This result provides novel insights into the next generation of magnetic devices based on complex materials and adds a new family of materials with positive magnetoresistance.

cond-mat.mtrl-sci↗

arXiv2021

Structural, electrical and energy storage properties of lead-free NaNbO3-BaHfO3 thin films

Lead-free dielectric thin-film capacitors with desirable energy storage density are gathering attention due to the increasing environmental concern and the integrating electronic devices. We here reported a series of new highly-orientated (1-x)NaNbO3-xBaHfO3 (x<0.15) lead-free thin films prepared by a sol-gel method, and presented the dependence of their structural, electrical and energy storage properties on the x level of BaHfO3. The microstructure, leakage current and breakdown strength of pristine NaNbO3 thin films are significantly improved by addition of BaHfO3. As a result, the superior energy storage performances were obtained at x=0.1 with recoverable energy storage density of 23.1 J/cm3 at 1100 kV/cm, excellent thermal stability from 30 to 210 0C, good fatigue resistance, and the fast charge-discharge rate.

cond-mat.mtrl-sci↗

arXiv2019

Cancellations of Resonances and Long Time Dynamics of Cubic Schrödinger Equation on $\mathbb{T}$

We prove a vanishing property of the normal form transformation of the 1D cubic nonlinear Schrödinger (NLS) equation with periodic boundary conditions on $[0,L]$. We apply this property to quintic resonance interactions and obtain a description of dynamics for time up to $T=\frac{L^2}{ε^4}$, if $L$ is sufficiently large and size of initial data $ε$ is small enough. Since $T$ is the characteristic time of wave turbulence, this result implies the absence of wave turbulence behavior of 1D cubic NLS. Our approach can be adapted to other integrable systems without too many difficulties. In the proof, we develop a correspondence between Feynman diagrams and terms in normal forms, which allows us to calculate the coefficients inductively.

math.AP↗

arXiv2017

Tuning the diffusion of magnon in Y3Fe5O12 by light excitation

Deliberate control of magnon transportation will lead to an energy-efficient technology for information transmission and processing. Y3Fe5O12(YIG), exhibiting extremely large magnon diffusion length due to the low magnetic damping constant, has been intensively investigated for decades. While most of the previous works focused on the determination of magnon diffusion length by various techniques, herein we demonstrated how to tune magnon diffusion by light excitation. We found that the diffusion length of thermal magnons is strongly dependent on light wavelength when the magnon is generated by exposing YIG directly to laser beam. The diffusion length, determined by a nonlocal geometry at room temperature, is ~30 um for the magnons produced by visible light (400-650 nm), and ~136-156 um for the laser between 808 nm and 980 nm. The diffusion distance is much longer than the reported value. In addition to thermal gradient, we found that light illumination affected the electron configuration of the Fe3+ ion in YIG. Long wavelength laser triggers a high spin to low spin state transition of the Fe3+ ions in FeO6 octahedron. This in turn causes a substantial softening of the magnon thus a dramatic increase in diffusion distance. The present work paves the way towards an efficient tuning of magnon transport behavior which is crucially important for magnon spintronics.

cond-mat.mtrl-sci↗