Effectively managing missing modalities is a fundamental challenge in real-world multimodal learning scenarios, where data incompleteness often results from systematic collection errors or sensor failures. Sparse Mixture-of-Experts (SMoE) architectures have the potential to naturally handle multimodal data, with individual experts specializing in different modalities. However, existing SMoE approach often lacks proper ability to handle missing modality, leading to performance degradation and poor generalization in real-world applications. We propose ConfSMoE to introduce a two-stage imputation module to handle the missing modality problem for the SMoE architecture by taking the opinion of experts and reveal the insight of expert collapse from theoretical analysis with strong empirical evidence. Inspired by our theoretical analysis, ConfSMoE propose a novel expert gating mechanism by detaching the softmax routing score to task confidence score w.r.t ground truth signal. This naturally relieves expert collapse without introducing additional load balance loss function. We show that the insights of expert collapse aligns with other gating mechanism such as Gaussian and Laplacian gate. The proposed method is evaluated on four different real world dataset with three distinct experiment settings to conduct comprehensive analysis of ConfSMoE on resistance to missing modality and the impacts of proposed gating mechanism.

Hopfield networks using Hebbian learning suffer from limited storage capacity. While supervised methods like Linear Logistic Regression (LLR) offer some improvement, kernel methods like Kernel Logistic Regression (KLR) significantly enhance storage capacity and noise robustness. However, KLR requires computationally expensive iterative learning. We propose Kernel Ridge Regression (KRR) as an efficient kernel-based alternative for learning high-capacity Hopfield networks. KRR utilizes the kernel trick and predicts bipolar states via regression, crucially offering a non-iterative, closed-form solution for learning dual variables. We evaluate KRR and compare its performance against Hebbian, LLR, and KLR. Our results demonstrate that KRR achieves state-of-the-art storage capacity (reaching a storage load of 1.5) and noise robustness, comparable to KLR. Crucially, KRR drastically reduces training time, being orders of magnitude faster than LLR and significantly faster than KLR, especially at higher storage loads. This establishes KRR as a potent and highly efficient method for building high-performance associative memories, providing comparable performance to KLR with substantial training speed advantages. This work provides the first empirical comparison between KRR and KLR in the context of Hopfield network learning.

Deep Gaussian Processes (DGPs) are powerful surrogate models known for their flexibility and ability to capture complex functions. However, extending them to multi-output settings remains challenging due to the need for efficient dependency modeling. We propose the Deep Intrinsic Coregionalization Multi-Output Gaussian Process (deepICMGP) surrogate for computer simulation experiments involving multiple outputs, which extends the Intrinsic Coregionalization Model (ICM) by introducing hierarchical coregionalization structures across layers. This enables deepICMGP to effectively model nonlinear and structured dependencies between multiple outputs, addressing key limitations of traditional multi-output GPs. We benchmark deepICMGP against state-of-the-art models, demonstrating its competitive performance. Furthermore, we incorporate active learning strategies into deepICMGP to optimize sequential design tasks, enhancing its ability to efficiently select informative input locations for multi-output systems.

Despite the widespread adoption of deterministic samplers in diffusion models (DMs), their potential limitations remain largely unexplored. In this paper, we identify collapse errors, a previously unrecognized phenomenon in ODE-based diffusion sampling, where the sampled data is overly concentrated in local data space. To quantify this effect, we introduce a novel metric and demonstrate that collapse errors occur across a variety of settings. When investigating its underlying causes, we observe a see-saw effect, where score learning in low noise regimes adversely impacts the one in high noise regimes. This misfitting in high noise regimes, coupled with the dynamics of deterministic samplers, ultimately causes collapse errors. Guided by these insights, we apply existing techniques from sampling, training, and architecture to empirically support our explanation of collapse errors. This work provides intensive empirical evidence of collapse errors in ODE-based diffusion sampling, emphasizing the need for further research into the interplay between score learning and deterministic sampling, an overlooked yet fundamental aspect of diffusion models.

We consider contextual bandit learning under distribution shift when reward vectors are ordered according to a given preference cone. We propose an adaptive-discretization and optimistic elimination based policy that self-tunes to the underlying distribution shift. To measure the performance of this policy, we introduce the notion of preference-based regret which measures the performance of a policy in terms of distance between Pareto fronts. We study the performance of this policy by establishing upper bounds on its regret under various assumptions on the nature of distribution shift. Our regret bounds generalize known results for the existing case of no distribution shift and vectorial reward settings, and scale gracefully with problem parameters in presence of distribution shifts.

This paper claims that machine learning models deployed in high stakes domains such as medicine must be interpretable, shareable, reproducible and accountable. We argue that these principles should form the foundational design criteria for machine learning algorithms dealing with critical medical data, including survival analysis and risk prediction tasks. Black box models, while often highly accurate, struggle to gain trust and regulatory approval in health care due to a lack of transparency. We discuss how intrinsically interpretable modeling approaches (such as kernel methods with sparsity, prototype-based learning, and deep kernel models) can serve as powerful alternatives to opaque deep networks, providing insight into biomedical predictions. We then examine accountability in model development, calling for rigorous evaluation, fairness, and uncertainty quantification to ensure models reliably support clinical decisions. Finally, we explore how generative AI and collaborative learning paradigms (such as federated learning and diffusion-based data synthesis) enable reproducible research and cross-institutional integration of heterogeneous biomedical data without compromising privacy, hence shareability. By rethinking machine learning foundations along these axes, we can develop medical AI that is not only accurate but also transparent, trustworthy, and translatable to real-world clinical settings.

We present a comprehensive theoretical analysis of first-order methods for escaping strict saddle points in smooth non-convex optimization. Our main contribution is a Perturbed Saddle-escape Descent (PSD) algorithm with fully explicit constants and a rigorous separation between gradient-descent and saddle-escape phases. For a function $f:\mathbb{R}^d\to\mathbb{R}$ with $\ell$-Lipschitz gradient and $ρ$-Lipschitz Hessian, we prove that PSD finds an $(ε,\sqrt{ρε})$-approximate second-order stationary point with high probability using at most $O(\ellΔ_f/ε^2)$ gradient evaluations for the descent phase plus $O((\ell/\sqrt{ρε})\log(d/δ))$ evaluations per escape episode, with at most $O(\ellΔ_f/ε^2)$ episodes needed. We validate our theoretical predictions through extensive experiments across both synthetic functions and practical machine learning tasks, confirming the logarithmic dimension dependence and the predicted per-episode function decrease. We also provide complete algorithmic specifications including a finite-difference variant (PSD-Probe) and a stochastic extension (PSGD) with robust mini-batch sizing.

In this paper, we propose a new numerical method for the underdamped Langevin diffusion (ULD) and present a non-asymptotic analysis of its sampling error in the 2-Wasserstein distance when the $d$-dimensional target distribution $p(x)\propto e^{-f(x)}$ is strongly log-concave and has varying degrees of smoothness. Precisely, under the assumptions that the gradient and Hessian of $f$ are Lipschitz continuous, our algorithm achieves a 2-Wasserstein error of $\varepsilon$ in $\mathcal{O}(\sqrt{d}/\varepsilon)$ and $\mathcal{O}(\sqrt{d}/\sqrt{\varepsilon})$ steps respectively. Therefore, our algorithm has a similar complexity as other popular Langevin MCMC algorithms under matching assumptions. However, if we additionally assume that the third derivative of $f$ is Lipschitz continuous, then our algorithm achieves a 2-Wasserstein error of $\varepsilon$ in $\mathcal{O}(\sqrt{d}/\varepsilon^{\frac{1}{3}})$ steps. To the best of our knowledge, this is the first gradient-only method for ULD with third order convergence. To support our theory, we perform Bayesian logistic regression across a range of real-world datasets, where our algorithm achieves competitive performance compared to an existing underdamped Langevin MCMC algorithm and the popular No U-Turn Sampler (NUTS).

--We propose a novel unsupervised outlier detection method based on Randomized Principal Component Analysis (PCA). Inspired by the performance of Randomized PCA (RPCA) Forest in approximate K-Nearest Neighbor (KNN) search, we develop a novel unsupervised outlier detection method that utilizes RPCA Forest for outlier detection. Experimental results showcase the superiority of the proposed approach compared to the classical and state-of-the-art methods in performing the outlier detection task on several datasets while performing competitively on the rest. The extensive analysis of the proposed method reflects it high generalization power and its computational efficiency, highlighting it as a good choice for unsupervised outlier detection. An outlier, as defined by Hawkins [18], is "an observation which deviates so much from other observations as to arouse suspicions that it was generated by a different mechanism." Similarly, Barnett and Lewis [3] describe it as "an observation (or subset of observations) which appears to be inconsistent with the remainder of that set of data." Outlier detection is the process of identifying such outliers, i.e., the data points which differ from the rest of the data. It is one of the most important and fundamental tasks in data mining and machine learning with applications in intrusion detection [20], fault detection [37], fraud detection [7] and others [11], [13], [27]. In recent years, many methods have been proposed to carry out the outlier detection task [1], [9], [10], [23], [42]. Despite the demonstration of promising results, further studies show that these results might be limited only to specific instances of the problem (e.g., a limited selection of datasets, a specific kind of outliers, etc.) [6].

The use of kernels for nonlinear prediction is widespread in machine learning. They have been popularized in support vector machines and used in kernel ridge regression, amongst others. Kernel methods share three aspects. First, instead of the original matrix of predictor variables or features, each observation is mapped into an enlarged feature space. Second, a ridge penalty term is used to shrink the coefficients on the features in the enlarged feature space. Third, the solution is not obtained in this enlarged feature space, but through solving a dual problem in the observation space. A major drawback in the present use of kernels is that the interpretation in terms of the original features is lost. In this paper, we argue that in the case of a wide matrix of features, where there are more features than observations, the kernel solution can be re-expressed in terms of a linear combination of the original matrix of features and a ridge penalty that involves a special metric. Consequently, the exact same predicted values can be obtained as a weighted linear combination of the features in the usual manner and thus can be interpreted. In the case where the number of features is less than the number of observations, we discuss a least-squares approximation of the kernel matrix that still allows the interpretation in terms of a linear combination. It is shown that these results hold for any function of a linear combination that minimizes the coefficients and has a ridge penalty on these coefficients, such as in kernel logistic regression and kernel Poisson regression. This work makes a contribution to interpretable artificial intelligence.