Kullback-Leibler Divergence

Also known as KL

Latest papers

31 papers4 papers in the last four weeks, down 43% on the four weeks before. 0.0% of all new papers.

Nov 13, 2026

On numerical issues of divergences between nearly equal distributions

Barron et al. (2026) recently identified numerical issues in the softmax and soft cross-entropy loss. We show that similar issues are present in standard divergence operations between categorical distributions. This includes the forward and reverse Kullback-Leibler (KL) divergence and generalized Jensen-Shannon divergence (JSD), as well as Monte Carlo KL estimates and entropy. When the distributions are nearly equal, rounding and cancellation issues affect the forward and gradient computations. These issues can affect settings using divergences in optimization objectives, such as knowledge distillation, reinforcement learning, and proximal methods. We propose numerically stable reformulations and provide example implementations for each setting, validated by property-based tests.
Nov 13, 2026

Lipschitz-Constrained Discriminators in Generative Modeling: Asymptotic Behavior and Support Mismatch

We study the asymptotic behavior of optimal discriminators associated to L−L-Lipschitz regularized KL-divergences of two probability distributions QQ and PP that interpolate between the rescaled Wasserstein−1-1 metric by some parameter L>0L>0 and the KL divergence. The parameter LL plays the role of a Lipschitz constraint and is viewed as a regularization parameter: small values of LL enforce a more geometry-aware discriminator, while as LL increases the geometric regularization is progressively relaxed. It is a known result [3] that such divergences converge to the KL divergence as L→∞L\to\infty. In this paper, we characterize the asymptotic behavior of the optimal discriminators as L→∞L\to\infty in different regimes: Firstly, under Q≪PQ\ll P and the Radon-Nikodym derivative dQdP\frac{dQ}{dP} is bounded from above and bounded away from zero. Secondly, when QQ and PP are discrete distributions satisfying Q≪PQ\ll P, and finally when QQ and PP are non absolutely continuous discrete distributions. This result motivates training strategies which involve discriminators, such as adversarial generative models. In practice, the data distribution is often represented by a discrete empirical measure, while, especially at the early stages of the training, the model distribution may have little or no overlap with the empirical data distribution.
Nov 3, 2026

Support Vector Machines Directly Minimise KL Divergence

Support vector machines (SVMs) are fundamentally grounded in geometric margin maximisation and currently lack a mainstream probabilistic interpretation. This keeps them theoretically isolated from probabilistic machine learning models that are naturally comparable and combinable. In this paper, we bridge this gap by proving that the dual objectives of both support vector classification and support vector regression are exact, constrained Kullback-Leibler (KL) divergence minimisation tasks between two Gaussian measures. We further show that this same divergence formulation recovers the exact objective function for kernel ridge regression when domain constraints are removed and L2L^2 regularisation is applied. By demonstrating that SVM duals directly minimise relative entropy, our results provide a unified information-theoretic model for standard kernel machines and lay the foundation for unifying broader kernel-based algorithms under probability theory.
Oct 27, 2026

Displacement Moments and Path-Terminal Separation in Brownian Interpolant Sampling

We prove early-stopped, target-to-output KL guarantees for Brownian interpolant Euler sampling from the true source under arbitrary Borel couplings, even with nonintegrable endpoints. At fixed stopping time and noise level, a fourth displacement moment yields an O(N−1)O(N^{-1}) discretization term. A second moment suffices for fixed-law convergence with grid refinement and vanishing drift error. For one fixed coupling with no displacement moment above two, the original and comparison paths have KL of orders one and 1/log⁡N1/\log N relative to the exact-drift Euler path. On the same geometric grid, terminal KL is at most C/N2C/N^2. The path decomposition identifies the information about the source discarded by Markov projection. When the original bridge is Markov, current-state Euler sampling has an O((M2+d)/N)O((M_2+d)/N) discretization term. We prove expected stopped-output KL consistency for a learner trained on iid endpoint pairs with fresh true-source initialization. In a spatial image control, iterative sampling has worse fair energy scores than direct prediction with resampled residuals. Uniform terminal rates for general second-moment couplings remain open.
Oct 16, 2026

Information Estimation with Discrete Diffusion

Information-theoretic measures, such as Mutual Information (MI), play a crucial role in understanding non-linear relationships between random variables and are widely used across scientific disciplines. Yet, their use on real-world discrete data remains challenging. Existing methods typically rely on embedding discrete data into a continuous space and apply neural estimators originally designed for continuous distributions. This process requires careful engineering for both the embedding model and estimator architecture, but suffers from issues related to high data dimensionality. In this work, we introduce InfoSEDD, a discrete diffusion-based approach that bridges information-theoretic estimation and generative modeling such that they can be used to compute Kullback-Leibler divergences. Backed by Continuous Time Markov Chains theory principles, the design of InfoSEDD is lightweight and scalable and allows seamless integration with pretrained models. We showcase the versatility of our approach through applications on motif discovery in genetic promoter data, semantic-aware model selection in text summarization, and entropy estimation in Ising models. Finally, we construct consistency tests on real-world textual and genomics data. Our experiments demonstrate that InfoSEDD outperforms alternatives that rely on the "embedding trick". Our results position InfoSEDD as a robust and scalable tool for information-theoretic analysis of discrete data.
Oct 14, 2026

Divergence-Routed On-Policy Distillation for Language Models

On-policy distillation (OPD) transfers knowledge from a teacher model to a student along student-generated trajectories, typically by minimizing the reverse Kullback-Leibler (RKL) divergence. However, RKL can provide weak corrective signals for teacher-supported tokens to which the student assigns very low probability. This raises two questions: which positions are most consequential for the behavioral gap between the student and teacher, and what additional supervision is needed at these positions? Through cross-sampling, we find that divergence-based routing more effectively pinpoints positions contributing to the teacher-student performance gap than entropy-based routing. Further analysis reveals teacher-mode under-coverage at high Jensen-Shannon (JS) divergence positions, where standard OPD increases the large teacher-student probability gaps. Motivated by these findings, we propose Divergence-Routed On-Policy Distillation (DrOPD). DrOPD retains RKL supervision at all positions and selectively adds forward Kullback-Leibler (FKL) divergence supervision at positions with high JS divergence. Experiments across six math benchmarks show that DrOPD improves over standard OPD and entropy-based routing. DrOPD also reduces severe under-coverage cases by 37.9 percent relative to OPD, supporting targeted FKL as an effective complement to OPD.
Oct 10, 2026

DisCo Algebra: Why Disintegration and Combination Are Not Inverses

We present DisCo algebra---a compositional framework for joint probability built from a symmetric cooperad (disintegration) and a magma-operad (non-associative partial combination, restricted to the image of disintegration) on the bare family Pn=Prob([0,1]n)\mathcal{P}_n=\mathrm{Prob}([0,1]^n). No probability integral transform, no external summary functions, no enrichment of objects with conditional kernels. The disintegration side uses block-wise joint CDFs as intrinsic, measure-determined 1D summaries, yielding a lossy compression. The combination side reads the same summaries and reconstructs via regular conditional measures, yielding a lossless embedding on admissible inputs. We prove that combination followed by disintegration is the identity on admissible inputs, but disintegration followed by combination is not---it defines a renormalization operator RI=Combine∘DisintegrateR_{\mathcal{I}}=\text{Combine}\circ\text{Disintegrate}. The information loss of this operator is exactly the probability curvature KI(μ)\mathcal{K}_{\mathcal{I}}(μ)---the conditional KL divergence between the true conditional distribution and the conditionally independent product. This curvature measures the obstruction to lossless decomposition and reveals a hard limit on how much cross-block dependence can be captured by CDF-based scalar block summaries.
Oct 9, 2026

KL Before and After Mixing: A Round-Complexity Separation in Parallel Sampling

A parallel sampler can choose at random which coordinates to generate together. For a target pp and the output law qUq_U of a schedule UU drawn afresh for each output, we compare the schedule-averaged divergence EUDKL(p∥qU)\mathbb{E}_U D_{\mathrm{KL}}(p\|q_U) with the output-law divergence DKL(p∥EUqU)D_{\mathrm{KL}}(p\|\mathbb{E}_U q_U), in nats. Rounds commit independent draws from exact conditional marginals frozen at the round's start, each output's schedule is drawn before sampling from a finite mixture of schedules, and every scheduled round is charged. For the uniform distribution on binary strings of even length nn with n/2n/2 ones, the least expected number of rounds at a fixed tolerance ε>0\varepsilon>0 is Θε((log⁡n)2)Θ_\varepsilon((\log n)^2) under the first criterion but (e−ε+o(1))log⁡2ln⁡n(e^{-\varepsilon}+o(1))\log_2\ln n under the second, and the double-logarithmic order is attainable with the same round count on every run. For this target, a count-state recursion gives the exact least output divergence at every round cap over all causal randomized policies. The separation extends to weighted fixed-count targets with bounded weight ratios and quota fractions bounded away from zero and one, without permutation symmetry, and to the full-support Beta(1,1)(1,1)--Bernoulli target, where the sharp expected cost under the output-law criterion is (e−ε/2+o(1))log⁡2log⁡2n(e^{-\varepsilon}/2+o(1))\log_2\log_2n, attained with rare nn-round runs. Prior work implies the order of the schedule-averaged cost on the balanced target and its Oε((log⁡n)2)O_\varepsilon((\log n)^2) upper bounds on both extensions.
Oct 9, 2026

Universal Bounds for Self-Normalized Kullback-Leibler Estimation

This work studies an entropy-based estimator of Kullback-Leibler (KL) divergence constructed from normalized importance weights. We prove a universal finite-sample bound showing that its expectation can exceed the target-proposal KL divergence by at most log⁡2\log 2 for any sample size. We then characterize the KL divergence between the distribution induced by importance-weighted resampling and the proposal, showing that it is upper bounded by both the expectation of the entropy-based estimator and the target-proposal KL divergence. Importantly, these bounds require neither bounded importance weights nor finite higher moments. We apply these results to test-time alignment of generative models via Soft Best-of-NN sampling, an importance-sampling-based scaling technique proposed by Verdun et al. (2025), and to relative entropy coding, obtaining finite-sample policy-divergence certificates and moment-free bounds on coding cost.
Oct 2, 2026

Divergence controls entropy in distillation

Distillation has become a core primitive of large language model training, but its properties are not yet well understood. We take an entropic perspective, studying how the entropy of the student depends on the data and the divergence that define the distillation objective. We prove that forward KL inflates the entropy of the student above that of the teacher. Since cross-entropy training is a special case, this yields an identity that we verify quantitatively in pretraining and supervised finetuning. Other divergences come with no such guarantee: reverse KL deflates entropy until the gap between student and teacher gets too large, and interpolating between the two changes entropy smoothly early in training but abruptly at convergence. The lower entropy of on-policy distillation comes from token-level reverse KL, not from on-policy sampling. The divergence therefore acts as an implicit entropy regularizer, whose role is clearest in self-distillation: as conditioning on privileged information deflates entropy, the divergence hyperparameters that work best are those that compensate for it.
Sep 30, 2026

How Divergence Becomes Decision Flips in Compressed Language Models

Compression reports summarize how far a compressed language model moved from the dense one, usually by a KL divergence; a deployment that relies on the dense model's outputs needs to know how many of its decisions changed. We show that total variation, not KL, answers this directly. Across 802 compressed and perturbed copies of 19 open models on five corpora and nine mechanically unrelated perturbation families, the rate at which the arg-max token changes (the flip rate) tracks total variation at a ratio with median 1.051.05, with no fitted constant. KL converts into flips only through its square root and a factor that varies fourfold across models and corpora, because KL averages over tokens before the root is taken; first-order statistics averaged per token, such as Hellinger distance, avoid this, but reports rarely give them. As a result, of two compressors reported on different models and corpora whose flip rates differ by at least 1010%, KL assigns the smaller divergence to the one that changes more decisions in 1111% of cases, total variation in 11%. Two pre-registered tests mark the limits: on a held-out code corpus the ratio held for all eight models while three predictions about KL each failed for half of them or more, and on three new models with real kernels it stayed in its band for 37 of 38 checkpoints but fell below one on code for two models. In vLLM speculative decoding, total variation measured under teacher forcing predicts greedy draft acceptance with a mean relative error of 1.11.1--2.42.4%, without the task-specific calibration that KL needs.
Sep 29, 2026

Understanding Off- vs On-Policy Distillation: A Tale of Distinct Training Objectives

On-policy distillation (OPD) learns from teacher feedback on student-generated responses and has shown promise in reducing forgetting relative to supervised fine-tuning (SFT). However, its benefits and fragility remain incompletely understood. We study sequential distillation from multiple teachers, where the student minimizes its average divergence from the teachers. Forward Kullback--Leibler (KL) divergence yields a weighted arithmetic mixture, while reverse KL yields a normalized weighted geometric aggregate. We develop algorithms that learn these targets under off-policy and on-policy feedback, respectively, establishing logarithmic regret bounds in the tabular setting and extending the analysis to function approximation. By analyzing these aggregation targets, we identify mechanisms that help explain both the benefits and fragility of OPD. Relative to forward KL, reverse KL can better retain a confident expert's preferences under uninformative feedback, but is more sensitive to teachers that assign very low probabilities to correct responses. Its token-level conditionals also reveal a dependence on continuation distributions that can favor incorrect prefixes over long horizons.
Sep 23, 2026

The Risk-Sensitive Schrödinger Bridge: Is Not a KL Projection

The Schrödinger bridge owes its computational power to a single structural fact: by Girsanov's theorem the controlled problem is a Kullback--Leibler (KL) projection onto a fixed reference measure, solvable by alternating projections. This letter shows that the fact does not survive risk sensitivity. When the expected path cost is replaced by the entropic risk measure and both endpoint marginals are kept as hard constraints, the resulting fixed-point bridge value JθJ_θ (the soft-problem value at the multiplier that enforces the terminal constraint) admits no representation as a constrained KL minimum against any fixed path-space reference with a regular endpoint law (a class strictly larger than the uniformly elliptic diffusion references: no Markov property is required), even allowing an additive normalisation depending on the initial marginal. Moreover, no single reference generates the one-parameter family in the risk parameter. The obstruction is computed in closed form: the Gaussian bridge value violates, by exactly θ/2θ/2, a heat equation that any Gaussian smoothing of a fixed endpoint density must obey. In place of the projection, the theory rests on a terminal-multiplier fixed point and an asymmetric factorisation penalising the score energy of the backward factor.
Sep 12, 2026

TF-IDF and BM25 Are Exact KL Divergences

TF-IDF and BM25 are two of the most widely used methods for scoring query-document relevance, yet neither has a standard probabilistic derivation that justifies it as a statistical method within a unified framework. We address this gap by showing that both scoring methods admit an exact interpretation as Kullback-Leibler divergences between two probability models. We treat the BM25 variant that includes the plus 1 correction in the IDF term, which is the one used in practice, and also discuss the original BM25 formulation without that correction. The resulting framework provides a common theoretical basis for TF-IDF and BM25, clarifies what they measure, and allows them to be compared theoretically with other information retrieval methods rather than only experimentally.
Sep 8, 2026

Mode Coverage in Normalizing Flow Boltzmann Generators via Log-Ratio Variation

Normalizing flow Boltzmann generators retain a tractable pushforward density, but training with forward KL depends on target samples that may be biased or omit modes. As a result, a flow can miss target mass while its observed importance weights give a high effective sample size. We introduce the log-ratio variation \Xω\X_ω, the mean absolute pairwise difference of the target-to-pushforward log-density ratio under a weighting measure ωω, and use it to define KLXX, a new loss function. Two log-ratio variations are added to the forward KL (denoted by the two X's): one weighted by the target to improve accuracy, the other by a mixture of quench and temper samples with pushforward samples to search candidate modes. We derive the Fisher--Rao gradient flow of KLXX, where both variations contribute nonpositive dissipation, and a fixed-surrogate error bound for KLXX. We use KLXX in an adaptive-staging Boltzmann generator, with importance reweighting at every stage. We bound the sampling error of its inference scheme when the stage weights are essentially bounded, and prove it asymptotically unbiased in the sample size. In the numerical tests, KLXX improves mode coverage over forward KL. It also improves the generator's per-stage diagnostics against the loss that built the schedule. The observables the generator recovers are close to independent references. The log-ratio variations thus supply information that the forward KL loss usually omits.
Aug 26, 2026

A Token-Level Analysis of Sampled-Token Reverse-KL On-Policy Distillation

On-policy distillation (OPD) supervises a student on its own trajectories with token-level signals from a frozen teacher, yet how a sampled loss allocates updates across tokens remains poorly understood. We analyze the gradient of the per-token K2 estimator of reverse KL with respect to the student logits. The ℓ1\ell_1 norm of this gradient factorizes into the absolute teacher--student log-probability gap and a student-side softmax factor that grows as the sampled token becomes less likely under the student. In our math-distillation runs, these per-token norms are highly non-uniform: low-student-probability tokens account for a disproportionate share of their sum and are also enriched in large teacher--student gaps. As a lightweight intervention suggested by this analysis, we study Surprise-aware Reweighting (SuRe), a detached, bounded weighting rule that further amplifies this existing allocation. Across two Qwen3 student scales, SuRe improves several math metrics over vanilla OPD and shows no clear degradation on the selected out-of-domain benchmarks. Our primary contribution is therefore a gradient-level characterization of reverse-KL OPD trained with the K2 estimator, with SuRe as one empirical instantiation.
Jul 5, 2026

Tightening the Score Matching Gap for Diffusion Models

Diffusion models (DMs) are a state-of-the-art generative method to approximately sample from an unknown distribution. Their training and evaluation primarily rely on an Evidence Lower Bound (ELBO), which relates the Kullback-Leibler (KL) divergence of model samples to the score matching loss along the path, which serves as a tractable surrogate. The difference between sample quality and the score matching loss produced by this bound leads to the score matching gap, which is known to be tight in the worst-case but not descriptive of sample quality in general. In this work, we provide a theoretical analysis of this gap, developing tighter bounds for three metrics: KL divergence, reverse KL divergence, and Wasserstein distance, effectively exploiting the regularity of the class of score estimators. Our results suggest that the quality of the score approximation has more impact on closing the score matching gap for low noise scales. To obtain these bounds, our key technical insight is to exploit the contraction properties of the backward processes. In particular, we rely on entropy flows, logarithmic Sobolev inequalities and reflection couplings, rigorously linking the ergodicity of the Langevin diffusion to the score matching gap problem.
Jun 25, 2026

Beyond Global Divergences: A Local-Mass Perspective on Bayesian Inference

Global objectives, such as KL divergence and ELBO, are widely used in Bayesian inference for measuring distributional discrepancy. This paper studies their local-mass behaviour that is not directly captured by such objectives. We introduce and use two mathematical tools: (1) Mass Index for recording the polynomial and logarithmic decay scales of local mass, and (2) regularised extended KL (RE-KL), a set-localised divergence that can be formulated in the presence of singular components. Mass Indices help characterise how Bayesian updating changes local mass: (1) power-log likelihood factors shift it explicitly, and (2) parameter-dependent supports, or their smooth softenings, may change the local scale through the amount of mass that remains near the parameter value. Using local RE-KL, we prove absolute, relative, and directional inequalities for comparing local small-ball masses under the two KL directions. Together, these results provide a local theoretical account of local mass behaviour. Experiments provide controlled illustrations of the local behaviour. Code is available at https://github.com/Forsythia0604/Local-Mass-Framework.
Jun 17, 2026

Displacement Is Not Direction: Evaluating Fidelity Metrics for Quantized LLM Deployment

Fidelity metrics, such as per-token KL divergence (KLD) against a high-precision reference, are often used in practice as low-cost proxies for benchmark quality. We test this practice on a 28-quant cohort of Qwen3.6-35B-A3B and a 41-quant cohort of Devstral-Small-2-24B, evaluated across a suite of downstream benchmarks. We find that KLD is strongly correlated with benchmark score over the full cohort (ρ=−0.72ρ=-0.72 on Qwen and ρ=−0.86ρ=-0.86 on Devstral, both with p<0.001p<0.001). However, this relationship collapses to non-significance in the near-baseline silent zone (ρ=+0.00ρ=+0.00 on Qwen and ρ=−0.24ρ=-0.24, p=0.36p=0.36, on Devstral). This collapse persists across 14 measurement variants, including different KLD aggregations, perplexity formulations, top-1 agreement, calibration corpora, and context lengths. At the per-prompt level, KLD has only weak failure-prediction power on code, with failed-vs-passed geometric-mean ratios in [1.08,1.22][1.08,1.22] across five models on LiveCodeBench, and fails as a cross-model router, achieving only 42.3%−49.4%42.3\%-49.4\% accuracy on disagreement prompts. We trace the collapse to a structural decomposition: KLD primarily measures the volume of disagreement with the reference, with silent-zone composite ρ=+0.94ρ=+0.94 (p<0.001p<0.001) on Qwen and +0.55+0.55 (p=0.03p=0.03) on Devstral, while its relationship to the direction of those disagreements is weak and task-conditional.
Jun 15, 2026

Diffusion Flow Matching: Dimension-Improved KL Bounds and Wasserstein Guarantees

Diffusion Flow Matching (DFM) has recently emerged as a versatile framework for generative modeling, yet its theoretical convergence properties remain only partially understood. In this work, we provide refined and novel convergence guarantees for Brownian motion based DFMs, focusing on the discretization error. Our analysis is conducted under the Kullback-Leibler (KL) divergence and the 2-Wasserstein distance. Under finite-moment conditions and a mild score integrability assumption, we derive KL convergence bounds with improved dimensional dependence compared to prior work, achieving, up to our knowledge, state-of-the-art scaling under minimal conditions. We further extend the analysis to the 2-Wasserstein distance: under an additional first-order score integrability assumption and a weak log-concavity condition, we obtain convergence guarantees with dimensional dependence consistent with the KL case.
May 29, 2026

Rethinking the Role of Temperature in Large Language Model Distillation

Reverse Kullback-Leibler (RKL) divergence is widely favored over forward KL (FKL) in large language models (LLM) distillation, yet this preference is largely based on comparisons that omit the temperature ττ, overlooking its central role in softening teacher distributions and improving knowledge transfer. In this work, we revisit temperature in LLM distillation and show that it fundamentally changes the comparison between FKL and RKL. Our analysis reveals an asymmetric effect: temperature substantially enriches FKL with non-dominant token signals, whereas it mainly rescales RKL gradients, causing FKL to benefit much more from ττ scaling than RKL. This asymmetry overturns the standard empirical conclusion: although RKL outperforms FKL at τ=1τ=1, FKL consistently surpasses RKL at higher temperatures across instruction-following benchmarks. Moreover, the impact of temperature is not limited to FKL; it improves a broader family of distillation objectives, enabling simple KL-based methods to achieve competitive performance against recent state-of-the-art LLM distillation approaches.
May 11, 2026

A Spectral Framework for Closed-Form Relative Density Estimation

We propose a closed-form spectral framework for relative log-density estimation in linearly parameterized probabilistic models, including unnormalized and conditional models. This is achieved by representing the Kullback-Leibler (KL) divergence as an integral of weighted chi-squared divergences, converting KL estimation into a family of least-squares problems. We derive an explicit spectral formula based only on first- and second-order feature moments, yielding closed-form estimators of both divergences and log-density potentials for fixed features. The framework extends to a broad class of f-divergences and can be combined with kernelization or feature learning with neural networks. We prove convergence guarantees for the resulting estimators and empirically compare them on synthetic data with optimization-based variational formulations, including logistic and softmax regression for normalized conditional models.
Apr 30, 2026

Exponential families from a single KL identity

Exponential families encompass the distributions central to modern machine learning -- softmax, Gaussians, and Boltzmann distributions -- and underlie the theory of variational inference, entropy-regularized reinforcement learning, and RLHF. We isolate a simple identity for exponential families that expresses the KL difference KL(q∥pλ2)−KL(q∥pλ1)\mathrm{KL}(q \| p_{λ_2}) - \mathrm{KL}(q \| p_{λ_1}) in terms of the log-partition function A(λ)A(λ) and the moment μqμ_q. Remarkably, this identity together with the single fact that KL≥0\mathrm{KL} \geq 0 (with equality iff p=qp = q) suffices, by direct substitution and rearrangement, to derive a cluster of results that are classically obtained by separate, heavier arguments: a generalized three-point identity for arbitrary reference distributions, Pythagorean theorems for I-projections and reverse I-projections, convexity of the log-partition function, identification of its Legendre dual in KL terms, the Gibbs variational principle, and the explicit optimizer in KL-regularized reward maximization, including the exponential tilting formula underlying entropy-regularized control and RLHF. Beyond these purely algebraic consequences, standard analytic arguments recover the gradient formula for the log-partition function, the Bregman representation of within-family KL divergence, and the surjectivity of the moment map. The note is self-contained.
Apr 13, 2026

KL Divergence Between Gaussians: A Step-by-Step Derivation for the Variational Autoencoder Objective

Kullback-Leibler (KL) divergence is a fundamental concept in information theory that quantifies the discrepancy between two probability distributions. In the context of Variational Autoencoders (VAEs), it serves as a central regularization term, imposing structure on the latent space and thereby enabling the model to exhibit generative capabilities. In this work, we present a detailed derivation of the closed-form expression for the KL divergence between Gaussian distributions, a case of particular importance in practical VAE implementations. Starting from the general definition for continuous random variables, we derive the expression for the univariate case and extend it to the multivariate setting under the assumption of diagonal covariance. Finally, we discuss the interpretation of each term in the resulting expression and its impact on the training dynamics of the model.
Apr 13, 2026

Worst-Case Optimal ε\sqrtε-Rate Stability of KL Divergence under Perturbations of Gaussian Distributions

We establish a stability bound for the Kullback-Leibler (KL) divergence when its Gaussian reference distribution is perturbed. Existing relaxed triangle inequalities of KL divergence between Gaussian distributions require all three distributions to be Gaussian, which limits their use in applications such as out-of-distribution detection with flow-based generative models. Let N1\mathcal{N}_1 and N2\mathcal{N}_2 be dd-dimensional Gaussian distributions, and let PP be any dd-dimensional distribution satisfying EP[∥X∥2]<∞\mathbb{E}_P[\|X\|^2]<\infty and KL(P∥N1)<∞\mathrm{KL}(P\Vert \mathcal{N}_1)<\infty. If KL(N1∥N2)≤ε\mathrm{KL}(\mathcal{N}_1\Vert\mathcal{N}_2)\leqε for 0<ε<1/120<ε<1/12, we give an upper bound of ∣KL(P∥N1)−KL(P∥N2)∣\left|\mathrm{KL}(P\Vert\mathcal{N}_1)-\mathrm{KL}(P\Vert\mathcal{N}_2)\right|. Under suitable uniform boundedness conditions, the bound is 6dε2+Kε+o(ε) \frac{\sqrt{6dε}}{2}+K\sqrtε+o(\sqrtε) as ε→0ε\rightarrow 0 for fixed dd, where KK depends on the parameters of N1\mathcal{N}_1 and N2\mathcal{N}_2, and on EP[∥X∥]\mathbb{E}_P[\|X\|] and EP[∥X∥2]\mathbb{E}_P[\|X\|^2]. Furthermore, we prove that the ε\sqrtε dependence is worst-case optimal. The result extends stability guarantees restricted to Gaussian triples to comparisons involving an arbitrary distribution and two nearby Gaussian references. Experiments on synthetic distributions and latent representations from flow-based models support the predicted stability behavior.
Apr 8, 2026

Robust Multi-Agent Target Tracking in Intermittent Communication Environments via Analytical Belief Merging

Autonomous multi-agent target tracking in GPS-denied and communication-restricted environments (e.g., underwater exploration, subterranean search and rescue, and adversarial domains) forces agents to operate independently and only exchange information during brief reconnection windows. Because transmitting complete observation and trajectory histories is bandwidth-exhaustive, exchanging probabilistic belief maps serves as a highly efficient proxy that preserves the topology of agent knowledge. While minimizing divergence metrics to merge these decentralized beliefs is conceptually sound, traditional approaches often rely on numerical solvers that introduce critical quantization errors and artificial noise floors. In this paper, we formulate the decentralized belief merging problem as Forward and Reverse Kullback-Leibler (KL) divergence optimizations and derive their exact closed-form analytical solutions. By deploying these derivations, we mathematically eliminate optimization artifacts, achieving perfect mathematical fidelity while reducing the computational complexity of the belief merge to O(N∣S∣)\mathcal{O}(N|S|) scalar operations. Furthermore, we propose a novel spatially-aware visit-weighted KL merging strategy that dynamically weighs agent beliefs based on their physical visitation history. Validated across tens of thousands of distributed simulations, extensive sensitivity analysis demonstrates that our proposed method significantly suppresses sensor noise and outperforms standard analytical means in environments characterized by highly degraded sensors and prolonged communication intervals.
Mar 31, 2026

Diversity-Aware Reverse Kullback-Leibler Divergence for Large Language Model Distillation

Reverse Kullback-Leibler (RKL) divergence has recently emerged as the preferred objective for large language model (LLM) distillation, consistently outperforming forward KL (FKL), particularly in regimes with large vocabularies and significant teacher-student capacity mismatch, where RKL focuses learning on dominant modes rather than enforcing dense alignment. However, RKL introduces a structural limitation that drives the student toward overconfident predictions. We first provide an analysis of RKL by decomposing its gradients into target and non-target components, and show that non-target gradients consistently push the target logit upward even when the student already matches the teacher, thereby reducing output diversity. In addition, RKL provides weak supervision over non-target classes, leading to poor tail alignment. To address these issues, we propose Diversity-aware RKL (DRKL), which removes this gradient effect and strengthens non-target supervision while preserving the optimization benefits of RKL. Extensive experiments across datasets and model families demonstrate that DRKL consistently outperforms FKL, RKL, and other state-of-the-art distillation objectives, achieving better performance and a superior fidelity-diversity trade-off.
Mar 3, 2026

Surprisal-Rényi Free Energy

The forward and reverse Kullback-Leibler (KL) divergences arise as limiting objectives in learning and inference, yet they induce markedly different inductive biases that cannot be explained by expectation terms alone. We introduce the Surprisal-Rényi Free Energy (SRFE), a normalized Rényi-Chernoff log-moment functional of the likelihood ratio. Although SRFE is algebraically related to Rényi divergence for fixed values of the order parameter, its endpoint-symmetric normalization yields a distinct free-energy viewpoint for analyzing how forward- and reverse KL behavior emerges near singular limits. We show that SRFE recovers the forward and reverse KL divergences as endpoint limits and derive local expansions around both regimes in which the variance of the log-likelihood ratio appears as a first-order correction away from the KL objective. These expansions reveal an explicit mean-variance tradeoff governing departures from KL-dominated behavior. We further establish a Gibbs-type variational characterization in terms of weighted KL projections and prove that SRFE controls large deviations of excess code length through Chernoff-type bounds, giving a Minimum Description Length interpretation of the functional. We evaluate SRFE on two synthetic settings (two-dimensional Gaussian-mixture variational inference and multi-class classification) and the long-tailed CIFAR-10 and CIFAR-100 (CIFAR-LT) datasets. On CIFAR-LT, retraining the best configuration of each loss over five seeds, SRFE matches cross-entropy accuracy on CIFAR-10-LT and gives mixed results on CIFAR-100-LT, improving few-shot accuracy at one of three imbalance factors. Together, these results position SRFE as a variance- and tail-sensitive, normalized Rényi-Chernoff free energy that clarifies the geometric, statistical, and large-deviation structure underlying the transition between forward- and reverse-KL limits.
Mar 2, 2026

Dimension-Independent Convergence of Underdamped Langevin Monte Carlo in KL Divergence

Underdamped Langevin dynamics (ULD) is a widely-used sampler for Gibbs distributions π∝e−Vπ\propto e^{-V}, and is often empirically effective in high dimensions. However, existing non-asymptotic convergence guarantees for discretized ULD typically scale polynomially with the ambient dimension dd, leading to vacuous bounds when dd is large. The main known dimension-free result concerns the randomized midpoint discretization in Wasserstein-2 distance (Liu et al.,2023), while dimension-independent guarantees for ULD discretizations in KL divergence have remained open. We close this gap by proving the first dimension-free KL divergence bounds for discretized ULD. Our analysis refines the KL local error framework (Altschuler et al., 2025) to a dimension-free setting and yields bounds that depend on tr(H)\mathrm{tr}(\mathbf{H}), where H\mathbf{H} upper bounds the Hessian of VV, rather than on dd. As a consequence, we obtain improved iteration complexity for underdamped Langevin Monte Carlo relative to overdamped Langevin methods in regimes where tr(H)≪d\mathrm{tr}(\mathbf{H})\ll d.
Jan 20, 2026

A Gaussian Perspective for Distributional Discrepancy in Generative Diffusion Models

This paper introduces an analytical approach to quantifying and optimizing the distributional discrepancy in generative diffusion models. For a multivariate Gaussian source, we explicitly derive the closed-form evolution trajectory and the resulting Kullback-Leibler (KL) divergence between the distributions of the source data and the reversely sampled data. Asymptotic analysis via the Euler-Maclaurin expansion characterizes the convergence behavior of this KL divergence, extracting its dominant term as an explicit functional of the noise schedule. Minimizing this dominant term via the calculus of variations yields a noise schedule described by a tangent law, inherently determined by the source covariance spectrum. We further prove that the Gaussian source exhibits an extremal property for the KL divergence among general source distributions with a given covariance. We also utilize the analytical KL divergence as a principled metric to identify efficient time discretization strategies for pretrained diffusion models, and demonstrate via experiments over diverse datasets that the identified strategies consistently outperform established baselines, particularly under constrained function evaluation budgets.