It remains difficult to evaluate machine learning classifiers in the absence of a large, labeled dataset. While labeled data can be prohibitively expensive or impossible to obtain, unlabeled data is plentiful. Here, we introduce Semi-Supervised Model Evaluation (SSME), a method that uses both labeled and unlabeled data to evaluate machine learning classifiers. SSME is the first evaluation method to take advantage of the fact that: (i) there are frequently multiple classifiers for the same task, (ii) continuous classifier scores are often available for all classes, and (iii) unlabeled data is often far more plentiful than labeled data. The key idea is to use a semi-supervised mixture model to estimate the joint distribution of ground truth labels and classifier predictions. We can then use this model to estimate any metric that is a function of classifier scores and ground truth labels (e.g., accuracy or expected calibration error). We present experiments in four domains where obtaining large labeled datasets is often impractical: (1) healthcare, (2) content moderation, (3) molecular property prediction, and (4) image annotation. Our results demonstrate that SSME estimates performance more accurately than do competing methods, reducing error by 5.1 relative to using labeled data alone and 2.4 relative to the next best competing method. SSME also improves accuracy when evaluating performance across subsets of the test distribution (e.g., specific demographic subgroups) and when evaluating the performance of language models. Rigorous evaluation is essential to the safe deployment of machine learning classifiers. The standard approach is to measure classifier performance using a large labeled dataset. In practice, however, labeled data is often scarce (Culotta & McCallum, 2005; Dutta & Das, 2023). Exacerbating the challenge of evaluation, the number of off-the-shelf classifiers has increased dramatically through the widespread usage of model hubs. The modern machine learning practitioner thus has a myriad of trained models, but little labeled data with which to evaluate them. In many domains, unlabeled data is much more abundant than labeled data (Bepler et al., 2019; Sagawa et al., 2021; Movva et al., 2024).

The author responses answered my questions as well as points raised by other reviewers, providing additional clarification.] This paper formulates a fully probabilistic model of the vesicle-release dynamics at the sub-cellular biophysical level in the ribbon synapse. The paper then develops a likelihood-free inference method, tests it on a synthetic dataset, and finally infers the parameters of vesicle release in the ribbon synapse from real data. Originality: The paper presents a novel combination of biophysical modeling of ribbon synapse and a likelihood-free inference of the parameters. To my knowledge, the fully stochastic modeling of the vesicle-release dynamics is itself new.

This is an interesting paper on a mechanistic model of the ribbon synapse along with an ABC inference approach. Neither component is particularly novel, but the paper is thorough and compelling. The audience will likely be computationally-savvy experimental neuroscientists and those interested in applications of ABC; the former may be harder to find at NeurIPS, though they do exist. I encourage the authors to make the suggested revisions before the camera ready deadline.

Inferring the true demand for a product or a service from aggregate data is often challenging due to the limited available supply, thus resulting in observations that are censored and correspond to the realized demand, thereby not accounting for the unsatisfied demand. Censored regression models are able to account for the effect of censoring due to the limited supply, but they don't consider the effect of substitutions, which may cause the demand for similar alternative products or services to increase. This paper proposes Diffusion-aware Censored Demand Models, which combine a Tobit likelihood with a graph diffusion process in order to model the latent process of transfer of unsatisfied demand between similar products or services. We instantiate this new class of models under the framework of GPs and, based on both simulated and real-world data for modeling sales, bike-sharing demand, and EV charging demand, demonstrate its ability to better recover the true demand and produce more accurate out-of-sample predictions.

Model uncertainty quantification involves measuring and evaluating the uncertainty linked to a model's predictions, helping assess their reliability and confidence. Noise injection is a technique used to enhance the robustness of neural networks by introducing randomness. In this paper, we establish a connection between noise injection and uncertainty quantification from a Bayesian standpoint. We theoretically demonstrate that injecting noise into the weights of a neural network is equivalent to Bayesian inference on a deep Gaussian process. Consequently, we introduce a Monte Carlo Noise Injection (MCNI) method, which involves injecting noise into the parameters during training and performing multiple forward propagations during inference to estimate the uncertainty of the prediction. Through simulation and experiments on regression and classification tasks, our method demonstrates superior performance compared to the baseline model.

I think this paper addresses an important issue and makes valuable contributions, and thus should be published. I have a few concerns, hence my lower rating for the last question above (which I think could be addressed relatively easily, however). I think this is fundamentally *OK* and even perhaps a positive thing. However, I think a bit more discussion needs to be given to how the arguments might be made more formal. For example, in Section 2.1, I think the proof is intended to hold only in the limit of M going to infinity. Please give a stament of what should hold in what limit-- this wasn't clear to me.

The proposed method is very similar to previous work by Nie et al. -- both use k-trees to search for low-treewidth Bayesian networks, both start with a randomly chosen initial clique, and both propose using an A* method for finding the best tree. The differences are that Nie et al. score k-trees using a mutual information score and use BDeu for choosing the final consistent Bayesian network, while this paper proposes using BIC and incrementally building the Bayesian network along with the k-tree, using the BN to score the k-tree. This paper also includes the additional restriction that the complete variable (partial) order is chosen randomly, while in Nie et al. The main justification for these differences is the ability to scale to large treewidths. However, in the experiments, the previous S2 algorithm also can scale to large treewidths.

This paper provides an interesting connection between kernel Bayesian inference and vector valued regression. Based on this, a new regularization method is provided to compute an approximation of the kernel embedding of the posterior distribution. Simulation results look promising, suggesting that the new method gains improvement over many existing methods. However, as a non expert, from reading the current introduction, I'm still confused about the motivation of using kernel Bayesian inference---in order to approximate the kernel embedding of the posterior, a sample of iid draws (x_i, y_i) from the joint distribution of the parameter/hidden variable (X in the paper) and data (Y in the paper) are assumed to be available. First, it is a highly non-trivial problem of obtaining samples (x_i)'s from the posterior.

Overall, I found the paper interesting; the paper offers new theory as well as numerical results comparable to the state of the art on decently difficult datasets. Perhaps due to space constraints, an important part of the paper (section 3.2) - the inference algorithm - is poorly explained. In particular, I initially thought that the use of particles meant that the approximating distribution was a sum of Dirac delta functions - but that cannot be the case since, even with many particles, the'posterior' would degenerate into the MAP (note that in similar work, authors either use particles when p(x) involves discrete x variables, as in Kulkarni et al, or'smooth' the particles to approximate a continuous distribution, as in Gershman et al). Instead, it looks like the algorithm works directly on samples of the distribution q0, q1.. (hence the vague'for whatever distribution q that {xi}ni 1 currently represents'). It is tempting to consider q_i to be a kernel density estimate (mixture of normals with fixed width), and see if we can approximate equation 9 for that representation to be stable.

The paper explains well how computing RSA using estimates of regression weights can result in a biased similarity matrix. However, in many cases in neuroscience, the RSA is computed directly on the patterns of activity, and not the estimates of regression weights beta. This diminishes the relevance of this paper to the neuroscience field. The authors very briefly address this alternate way of computing RSA in lines 123-128. It is unclear how this alternative RSA computation is biased if it does not depend on a proxy for beta estimates, and needs to be addressed further.