arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2511.21076cs.LG
收藏

Deceptron:学习局部反向用于快速稳定物理反向

Deceptron: Learned Local Inverses for Fast and Stable Physics Inversion

Independent Researcher(独立研究者)

Aaditya L. Kachhadiya

AI总结:

Deceptron通过学习局部反向来加速和稳定物理反向问题,其D-IPG方法在Heat-1D和阻尼振子问题中表现出更高的效率,同时保持与高斯-牛顿方法相当的迭代次数和成本。

摘要:

在物理科学中的反问题通常在输入空间中是病态的,使得进展对步长敏感。我们提出了Deceptron,一个轻量级的双向模块,它学习可微前向替代的局部反向。训练结合了监督拟合、正反一致性、轻量级谱惩罚、软偏置绑定以及一个鼓励$J_g(f(x))\,J_f(x)\!\approx\!I$的雅可比乘积惩罚(JCP)通过JVP/VJP探测。在求解时,D-IPG(Deceptron反向预条件梯度)在输出空间中进行下降步,将其拉回通过$g$,并在与基线相同的回溯和停止规则下进行投影。在Heat-1D初始条件恢复和阻尼振子反问题上,D-IPG在Heat上达到固定归一化容差所需的迭代次数比投影梯度少约20倍,在Oscillator上少约2-3倍,与高斯-牛顿在迭代次数和成本上具有竞争力。诊断显示JCP减少了测量的乘积误差并跟踪迭代收益。我们还预览了一个单尺度2D实例,DeceptronNet(v0),它在严格的公平协议下学习少量步骤的修正,并显示出显著快的收敛速度。

英文摘要:

Inverse problems in the physical sciences are often ill-conditioned in input space, making progress step-size sensitive. We propose the Deceptron, a lightweight bidirectional module that learns a local inverse of a differentiable forward surrogate. Training combines a supervised fit, forward-reverse consistency, a lightweight spectral penalty, a soft bias tie, and a Jacobian Composition Penalty (JCP) that encourages $J_g(f(x))\,J_f(x)\!\approx\!I$ via JVP/VJP probes. At solve time, D-IPG (Deceptron Inverse-Preconditioned Gradient) takes a descent step in output space, pulls it back through $g$, and projects under the same backtracking and stopping rules as baselines. On Heat-1D initial-condition recovery and a Damped Oscillator inverse problem, D-IPG reaches a fixed normalized tolerance with $\sim$20$\times$ fewer iterations on Heat and $\sim$2-3$\times$ fewer on Oscillator than projected gradient, competitive in iterations and cost with Gauss-Newton. Diagnostics show JCP reduces a measured composition error and tracks iteration gains. We also preview a single-scale 2D instantiation, DeceptronNet (v0), that learns few-step corrections under a strict fairness protocol and exhibits notably fast convergence.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑