Causal self-attention is a coupling mechanism: each token's hidden state is updated by a learned mixture of preceding tokens at the same timescale. This paper asks whether a second, temporally slower coupling-a slow sub-system operating on a temporally-downsampled view of the sequence and fed back into the fast path through a zero-initialised gate-complements it. The question is framed in the language of singularly perturbed ordinary differential equations (ODEs), where the fast variable $x$ evolves at the token rate, the slow variable $y$ evolves at one update per $P$ tokens, and the timescale ratio $\varepsilon = 1/P$ is enforced structurally by causal block-mean pooling. The paper instantiates the fast-slow ODE formalism as a concrete neural network: a fast path of standard causal attention over $T$ tokens, a slow path of full attention over $T/P$ pooled tokens ($P^2 \times$ cheaper per layer), and a zero-initialised additive gate. In addition, under a linear-generator assumption on the fast dynamics, we prove that the equilibrium manifold $x = ϕ(y)$ is exactly the master-equation (ME) stationary distribution $p_{\mathrm{st}}(y)$; in that regime a learned MLP $ϕ_θ(y)$ is a variational approximation of it (the trained block is not a generator, so this identity is the structured limit, not a claim about the network as trained). Empirically, at $500$k tokens the coupling is neutral -- the gate stays closed and the coupled and frozen ablations are within run-to-run noise -- at a wall-clock cost comparable to a dense baseline. The contribution is the precise, gap-marked mapping itself, not a performance gain.

This paper introduces stochastic recurrent neural networks which glue a deterministic recurrent neural network and a state space model together to form a stochastic and sequential neural generative model. The clear separation of deterministic and stochastic layers allows a structured variational inference network to track the factorization of the model's posterior distribution. By retaining both the nonlinear recursive structure of a recurrent neural network and averaging over the uncertainty in a latent path, like a state space model, we improve the state of the art results on the Blizzard and TIMIT speech modeling data sets by a large margin, while achieving comparable performances to competing methods on polyphonic music modeling.

Current black-box variational inference (BBVI) methods require the user to make numerous design choices--such as the selection of variational objective and approximating family--yet there is little principled guidance on how to do so. We develop a conceptual framework and set of experimental tools to understand the effects of these choices, which we leverage to propose best practices for maximizing posterior approximation accuracy. Our approach is based on studying the pre-asymptotic tail behavior of the density ratios between the joint distribution and the variational approximation, then exploiting insights and tools from the importance sampling literature. Our framework and supporting experiments help to distinguish between the behavior of BBVI methods for approximating low-dimensional versus moderate-to-high-dimensional posteriors. In the latter case, we show that mass-covering variational objectives are difficult to optimize and do not improve accuracy, but flexible variational families can improve accuracy and the effectiveness of importance sampling--at the cost of additional optimization challenges. Therefore, for moderate-to-high-dimensional posteriors we recommend using the (mode-seeking) exclusive KL divergence since it is the easiest to optimize, and improving the variational family or using model parameter transformations to make the posterior and optimal variational approximation more similar. On the other hand, in low-dimensional settings, we show that heavy-tailed variational families and mass-covering divergences are effective and can increase the chances that the approximation can be improved by importance sampling.

This research considers a scalable inference for spatial data modeled through Gaussian intrinsic conditional autoregressive (ICAR) structures. The classical estimation method, restricted maximum likelihood (REML), requires repeated inversion and factorization of large, sparse precision matrices, which makes this computation costly. To sort this problem out, we propose a variational restricted maximum likelihood (VREML) framework that approximates the intractable marginal likelihood using a Gaussian variational distribution. By constructing an evidence lower bound (ELBO) on the restricted likelihood, we derive a computationally efficient coordinate-ascent algorithm for jointly estimating the spatial random effects and variance components. In this article, we theoretically establish the monotone convergence of ELBO and mathematically exhibit that the variational family is exact under Gaussian ICAR settings, which is an indication of nullifying approximation error at the posterior level. We empirically establish the supremacy of our VREML over MLE and INLA.

We develop EigenVI, an eigenvalue-based approach for black-box variational inference (BBVI).

We perform experiments on high-dimensional Gaussians, Lotka-V olterra dynamics, and a Pickover attractor.

We develop EigenVI, an eigenvalue-based approach for black-box variational inference (BBVI).