Searcharxiv⌕ Search

arXiv subjects

Mina Rezaei

Publications and source records attributed to Mina Rezaei.

At least 19 recordsLinked to original sources
arXiv2026

Predicting the 2026 FIFA World Cup with Sufficient Dimension Reduction of Elo Rating Histories

We study probabilistic forecasting of the 2026 FIFA World Cup, the first edition with 48 teams and an added Round of 32. The main idea is to describe team strength not only by the current Elo rating, but by a short history of recent Elo differences. We then reduce this history to a few informative directions using categorical sufficient dimension reduction (SDR). The reduced scores are used in a Poisson double-regression model for home and away goals, which gives full outcome probabilities. We compare eleven models, including logistic regression, standard Poisson regression, ARIMA, and neural-network forecasts of the Elo series, gradient boosting, an ensemble model, and four categorical SDR variants based on sliced inverse regression (SIR) and sliced average variance estimation (SAVE). The models are evaluated out of sample on the 2018 and 2022 World Cups using the ranked probability score (RPS). The results show that SDR-based poisson models improve the traditional approaches, suggesting that recent Elo history contains useful predictive information that is not captured by the current Elo difference alone.

stat.AP↗
arXiv2026

What Intermediate Layers Know: Detecting Jailbreaks from Entropy Dynamics

Jailbreak attacks reveal a persistent weakness in aligned Large Language Models: carefully crafted prompts can elicit policy-violating responses despite safety training. While most defenses operate at the prompt or output level, it remains unclear how harmful intent is encoded within the model's internal representations. We investigate this question by analyzing token-level predictive entropy trajectories across layers of a frozen LLM using the logit lens. We find that static aggregate statistics of prompt-level entropy (e.g., mean, variance) carry little discriminative signal, whereas features capturing how entropy evolves across token positions, such as monotonic rank-based trend scores, are substantially more informative. Importantly, this signal is not uniform across model depth: it is concentrated in intermediate layers and degrades at the final layer, indicating that jailbreak-relevant structure is most pronounced in mid-network representations rather than at the output head. Across multiple models (Llama, Qwen, Gemma) and adversarial benchmarks, these entropy dynamics provide architecture-consistent separation without additional training. Together, our findings show that jailbreak behavior is reflected in structured intermediate uncertainty dynamics, clarifying both which entropy-derived features encode harmful intent and where in the network that signal is most pronounced.

cs.CL↗
arXiv2026

Structured Credal Learning

Real-world learning tasks often encounter uncertainty due to covariate shift and noisy or inconsistent labels. However, existing robust learning methods merge these effects into a single distributional uncertainty set. In this work, we introduce a novel structured credal learning framework that explicitly separates these two sources. Specifically, we derive geometric bounds on the total variation diameter of structured credal sets and demonstrate how this quantity decomposes into contributions from covariate shift and expected label disagreement. This decomposition reveals a gating effect: covariate modulates how much label disagreement contributes to the joint uncertainty, such that seemingly benign covariate shifts can substantially increase the effective uncertainty. We also establish finite-sample concentration bounds in a fixed covariate regime and demonstrate that this quantity can be efficiently estimated. Lastly, we show that robust optimization over these structured credal sets reduces to a tractable discrete min-max problem, avoiding ad-hoc robustness parameters. Overall, our approach provides a principled and practical foundation for robust learning under combined covariate and label mechanism ambiguity.

stat.ME↗
arXiv2025

Revisiting Theory of Contrastive Learning for Domain Generalization

Contrastive learning is among the most popular and powerful approaches for self-supervised representation learning, where the goal is to map semantically similar samples close together while separating dissimilar ones in the latent space. Existing theoretical methods assume that downstream task classes are drawn from the same latent class distribution used during the pretraining phase. However, in real-world settings, downstream tasks may not only exhibit distributional shifts within the same label space but also introduce new or broader label spaces, leading to domain generalization challenges. In this work, we introduce novel generalization bounds that explicitly account for both types of mismatch: domain shift and domain generalization. Specifically, we analyze scenarios where downstream tasks either (i) draw classes from the same latent class space but with shifted distributions, or (ii) involve new label spaces beyond those seen during pretraining. Our analysis reveals how the performance of contrastively learned representations depends on the statistical discrepancy between pretraining and downstream distributions. This extended perspective allows us to derive provable guarantees on the performance of learned representations on average classification tasks involving class distributions outside the pretraining latent class set.

stat.ML↗
arXiv2025

Mitigating Catastrophic Forgetting in Continual Learning through Model Growth

Catastrophic forgetting is a significant challenge in continual learning, in which a model loses prior knowledge when it is fine-tuned on new tasks. This problem is particularly critical for large language models (LLMs) undergoing continual learning, as retaining performance across diverse domains is important for their general utility. In this paper, we explore model growth, a promising strategy that leverages smaller models to expedite and structure the training of larger ones for mitigating the catastrophic forgetting problem. Although growth-based pretraining, particularly via transformer stacking, has shown promise in accelerating convergence, its impact on forgetting remains under-explored. Therefore, we evaluate whether growth-based models can retain previously learned capabilities more effectively across a sequence of fine-tuning tasks involving domain knowledge, reasoning, reading comprehension, and bias. Our findings show that both models -- one trained with growth (Stack LLM) and one without (LLM) -- exhibit improvements in domain knowledge. However, reasoning and reading comprehension degrade over time, indicating signs of catastrophic forgetting. Stack LLM consistently shows less degradation, especially in reading comprehension, suggesting enhanced retention capabilities. Interestingly, in bias evaluation, the baseline LLM becomes progressively more neutral with continued fine-tuning, while Stack LLM maintains a steady bias ratio around 60--61\%. These results indicate that growth-based pretraining may deliver modest improvements in resisting catastrophic forgetting, though trade-offs remain in handling social biases.

cs.CL↗
arXiv2025

crossMoDA Challenge: Evolution of Cross-Modality Domain Adaptation Techniques for Vestibular Schwannoma and Cochlea Segmentation from 2021 to 2023

The cross-Modality Domain Adaptation (crossMoDA) challenge series, initiated in 2021 in conjunction with the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), focuses on unsupervised cross-modality segmentation, learning from contrast-enhanced T1 (ceT1) and transferring to T2 MRI. The task is an extreme example of domain shift chosen to serve as a meaningful and illustrative benchmark. From a clinical application perspective, it aims to automate Vestibular Schwannoma (VS) and cochlea segmentation on T2 scans for more cost-effective VS management. Over time, the challenge objectives have evolved to enhance its clinical relevance. The challenge evolved from using single-institutional data and basic segmentation in 2021 to incorporating multi-institutional data and Koos grading in 2022, and by 2023, it included heterogeneous routine data and sub-segmentation of intra- and extra-meatal tumour components. In this work, we report the findings of the 2022 and 2023 editions and perform a retrospective analysis of the challenge progression over the years. The observations from the successive challenge contributions indicate that the number of outliers decreases with an expanding dataset. This is notable since the diversity of scanning protocols of the datasets concurrently increased. The winning approach of the 2023 edition reduced the number of outliers on the 2021 and 2022 testing data, demonstrating how increased data heterogeneity can enhance segmentation performance even on homogeneous data. However, the cochlea Dice score declined in 2023, likely due to the added complexity from tumour sub-annotations affecting overall segmentation performance. While progress is still needed for clinically acceptable VS segmentation, the plateauing performance suggests that a more challenging cross-modal task may better serve future benchmarking.

eess.IV↗
arXiv2025

Calibrating LLMs with Information-Theoretic Evidential Deep Learning

Fine-tuned large language models (LLMs) often exhibit overconfidence, particularly when trained on small datasets, resulting in poor calibration and inaccurate uncertainty estimates. Evidential Deep Learning (EDL), an uncertainty-aware approach, enables uncertainty estimation in a single forward pass, making it a promising method for calibrating fine-tuned LLMs. However, despite its computational efficiency, EDL is prone to overfitting, as its training objective can result in overly concentrated probability distributions. To mitigate this, we propose regularizing EDL by incorporating an information bottleneck (IB). Our approach IB-EDL suppresses spurious information in the evidence generated by the model and encourages truly predictive information to influence both the predictions and uncertainty estimates. Extensive experiments across various fine-tuned LLMs and tasks demonstrate that IB-EDL outperforms both existing EDL and non-EDL approaches. By improving the trustworthiness of LLMs, IB-EDL facilitates their broader adoption in domains requiring high levels of confidence calibration. Code is available at https://github.com/sandylaker/ib-edl.

cs.LG↗
arXiv2024

A Dual-Perspective Approach to Evaluating Feature Attribution Methods

Feature attribution methods attempt to explain neural network predictions by identifying relevant features. However, establishing a cohesive framework for assessing feature attribution remains a challenge. There are several views through which we can evaluate attributions. One principal lens is to observe the effect of perturbing attributed features on the model's behavior (i.e., faithfulness). While providing useful insights, existing faithfulness evaluations suffer from shortcomings that we reveal in this paper. In this work, we propose two new perspectives within the faithfulness paradigm that reveal intuitive properties: soundness and completeness. Soundness assesses the degree to which attributed features are truly predictive features, while completeness examines how well the resulting attribution reveals all the predictive features. The two perspectives are based on a firm mathematical foundation and provide quantitative metrics that are computable through efficient algorithms. We apply these metrics to mainstream attribution methods, offering a novel lens through which to analyze and compare feature attribution methods.

cs.LG↗
arXiv2024

FinerCut: Finer-grained Interpretable Layer Pruning for Large Language Models

Overparametrized transformer networks are the state-of-the-art architecture for Large Language Models (LLMs). However, such models contain billions of parameters making large compute a necessity, while raising environmental concerns. To address these issues, we propose FinerCut, a new form of fine-grained layer pruning, which in contrast to prior work at the transformer block level, considers all self-attention and feed-forward network (FFN) layers within blocks as individual pruning candidates. FinerCut prunes layers whose removal causes minimal alternation to the model's output -- contributing to a new, lean, interpretable, and task-agnostic pruning method. Tested across 9 benchmarks, our approach retains 90% performance of Llama3-8B with 25% layers removed, and 95% performance of Llama3-70B with 30% layers removed, all without fine-tuning or post-pruning reconstruction. Strikingly, we observe intriguing results with FinerCut: 42% (34 out of 80) of the self-attention layers in Llama3-70B can be removed while preserving 99% of its performance -- without additional fine-tuning after removal. Moreover, FinerCut provides a tool to inspect the types and locations of pruned layers, allowing to observe interesting pruning behaviors. For instance, we observe a preference for pruning self-attention layers, often at deeper consecutive decoder layers. We hope our insights inspire future efficient LLM architecture designs.

cs.LG↗
arXiv2024

Efficient Document Embeddings via Self-Contrastive Bregman Divergence Learning

Learning quality document embeddings is a fundamental problem in natural language processing (NLP), information retrieval (IR), recommendation systems, and search engines. Despite recent advances in the development of transformer-based models that produce sentence embeddings with self-contrastive learning, the encoding of long documents (Ks of words) is still challenging with respect to both efficiency and quality considerations. Therefore, we train Longfomer-based document encoders using a state-of-the-art unsupervised contrastive learning method (SimCSE). Further on, we complement the baseline method -- siamese neural network -- with additional convex neural networks based on functional Bregman divergence aiming to enhance the quality of the output document representations. We show that overall the combination of a self-contrastive siamese network and our proposed neural Bregman network outperforms the baselines in two linear classification settings on three long document topic classification tasks from the legal and biomedical domains.

cs.CL↗
arXiv2024

Uncertainty-aware Pseudo-label Selection for Positive-Unlabeled Learning

Positive-unlabeled learning (PUL) aims at learning a binary classifier from only positive and unlabeled training data. Even though real-world applications often involve imbalanced datasets where the majority of examples belong to one class, most contemporary approaches to PUL do not investigate performance in this setting, thus severely limiting their applicability in practice. In this work, we thus propose to tackle the issues of imbalanced datasets and model calibration in a PUL setting through an uncertainty-aware pseudo-labeling procedure (PUUPL): by boosting the signal from the minority class, pseudo-labeling expands the labeled dataset with new samples from the unlabeled set, while explicit uncertainty quantification prevents the emergence of harmful confirmation bias leading to increased predictive performance. Within a series of experiments, PUUPL yields substantial performance gains in highly imbalanced settings while also showing strong performance in balanced PU scenarios across recent baselines. We furthermore provide ablations and sensitivity analyses to shed light on PUUPL's several ingredients. Finally, a real-world application with an imbalanced dataset confirms the advantage of our approach.

stat.ML↗
arXiv2024

AttributionLab: Faithfulness of Feature Attribution Under Controllable Environments

Feature attribution explains neural network outputs by identifying relevant input features. The attribution has to be faithful, meaning that the attributed features must mirror the input features that influence the output. One recent trend to test faithfulness is to fit a model on designed data with known relevant features and then compare attributions with ground truth input features.This idea assumes that the model learns to use all and only these designed features, for which there is no guarantee. In this paper, we solve this issue by designing the network and manually setting its weights, along with designing data. The setup, AttributionLab, serves as a sanity check for faithfulness: If an attribution method is not faithful in a controlled environment, it can be unreliable in the wild. The environment is also a laboratory for controlled experiments by which we can analyze attribution methods and suggest improvements.

cs.LG↗
arXiv2023

Stochastic Vision Transformers with Wasserstein Distance-Aware Attention

Self-supervised learning is one of the most promising approaches to acquiring knowledge from limited labeled data. Despite the substantial advancements made in recent years, self-supervised models have posed a challenge to practitioners, as they do not readily provide insight into the model's confidence and uncertainty. Tackling this issue is no simple feat, primarily due to the complexity involved in implementing techniques that can make use of the latent representations learned during pre-training without relying on explicit labels. Motivated by this, we introduce a new stochastic vision transformer that integrates uncertainty and distance awareness into self-supervised learning (SSL) pipelines. Instead of the conventional deterministic vector embedding, our novel stochastic vision transformer encodes image patches into elliptical Gaussian distributional embeddings. Notably, the attention matrices of these stochastic representational embeddings are computed using Wasserstein distance-based attention, effectively capitalizing on the distributional nature of these embeddings. Additionally, we propose a regularization term based on Wasserstein distance for both pre-training and fine-tuning processes, thereby incorporating distance awareness into latent representations. We perform extensive experiments across different tasks such as in-distribution generalization, out-of-distribution detection, dataset corruption, semi-supervised settings, and transfer learning to other datasets and tasks. Our proposed method achieves superior accuracy and calibration, surpassing the self-supervised baseline in a wide range of experiments on a variety of datasets.

cs.CV↗
arXiv2023

Probabilistic Self-supervised Learning via Scoring Rules Minimization

In this paper, we propose a novel probabilistic self-supervised learning via Scoring Rule Minimization (ProSMIN), which leverages the power of probabilistic models to enhance representation quality and mitigate collapsing representations. Our proposed approach involves two neural networks; the online network and the target network, which collaborate and learn the diverse distribution of representations from each other through knowledge distillation. By presenting the input samples in two augmented formats, the online network is trained to predict the target network representation of the same sample under a different augmented view. The two networks are trained via our new loss function based on proper scoring rules. We provide a theoretical justification for ProSMIN's convergence, demonstrating the strict propriety of its modified scoring rule. This insight validates the method's optimization process and contributes to its robustness and effectiveness in improving representation quality. We evaluate our probabilistic model on various downstream tasks, such as in-distribution generalization, out-of-distribution detection, dataset corruption, low-shot learning, and transfer learning. Our method achieves superior accuracy and calibration, surpassing the self-supervised baseline in a wide range of experiments on large-scale datasets like ImageNet-O and ImageNet-C, ProSMIN demonstrates its scalability and real-world applicability.

cs.LG↗
arXiv2023

Diversified Ensemble of Independent Sub-Networks for Robust Self-Supervised Representation Learning

Ensembling a neural network is a widely recognized approach to enhance model performance, estimate uncertainty, and improve robustness in deep supervised learning. However, deep ensembles often come with high computational costs and memory demands. In addition, the efficiency of a deep ensemble is related to diversity among the ensemble members which is challenging for large, over-parameterized deep neural networks. Moreover, ensemble learning has not yet seen such widespread adoption, and it remains a challenging endeavor for self-supervised or unsupervised representation learning. Motivated by these challenges, we present a novel self-supervised training regime that leverages an ensemble of independent sub-networks, complemented by a new loss function designed to encourage diversity. Our method efficiently builds a sub-model ensemble with high diversity, leading to well-calibrated estimates of model uncertainty, all achieved with minimal computational overhead compared to traditional deep self-supervised ensembles. To evaluate the effectiveness of our approach, we conducted extensive experiments across various tasks, including in-distribution generalization, out-of-distribution detection, dataset corruption, and semi-supervised settings. The results demonstrate that our method significantly improves prediction reliability. Our approach not only achieves excellent accuracy but also enhances calibration, surpassing baseline performance across a wide range of self-supervised architectures in computer vision, natural language processing, and genomics data.

stat.ML↗
arXiv2023

Young Humans Make Change, Young Users Click: Creating Youth-Centered Networked Social Movements

From the urbanists' perspective, the everyday experience of young people, as an underrepresented group in the design of public spaces, includes tactics they use to challenge the strategies which rule over urban spaces. In this regard, youth led social movements are a set of collective tactics which groups of young people use to resist power structures. Social informational streams have revolutionized the way youth organize and mobilize for social movements throughout the world, especially in urban areas. However, just like public spaces, these algorithm based platforms have been developed with a great power imbalance between the developers and users which results in the creation of non inclusive social informational streams for young activists. Social activism grows agency and confidence in youth which is critical to their development. This paper employs a youth centric lens, which is used in designing public spaces, for designing algorithmic spaces that can improve bottom up youth led movements. By reviewing the structure of these spaces and how young people interact with these structures in the different cultural contexts of Iran and the US, we propose a humanistic approach to designing social informational streams which can enhance youth activism.

cs.HC↗
arXiv2022

FiLM-Ensemble: Probabilistic Deep Learning via Feature-wise Linear Modulation

The ability to estimate epistemic uncertainty is often crucial when deploying machine learning in the real world, but modern methods often produce overconfident, uncalibrated uncertainty predictions. A common approach to quantify epistemic uncertainty, usable across a wide class of prediction models, is to train a model ensemble. In a naive implementation, the ensemble approach has high computational cost and high memory demand. This challenges in particular modern deep learning, where even a single deep network is already demanding in terms of compute and memory, and has given rise to a number of attempts to emulate the model ensemble without actually instantiating separate ensemble members. We introduce FiLM-Ensemble, a deep, implicit ensemble method based on the concept of Feature-wise Linear Modulation (FiLM). That technique was originally developed for multi-task learning, with the aim of decoupling different tasks. We show that the idea can be extended to uncertainty quantification: by modulating the network activations of a single deep network with FiLM, one obtains a model ensemble with high diversity, and consequently well-calibrated estimates of epistemic uncertainty, with low computational overhead in comparison. Empirically, FiLM-Ensemble outperforms other implicit ensemble methods, and it and comes very close to the upper bound of an explicit ensemble of networks (sometimes even beating it), at a fraction of the memory cost.

cs.LG↗
arXiv2022

Joint Debiased Representation and Image Clustering Learning with Self-Supervision

Contrastive learning is among the most successful methods for visual representation learning, and its performance can be further improved by jointly performing clustering on the learned representations. However, existing methods for joint clustering and contrastive learning do not perform well on long-tailed data distributions, as majority classes overwhelm and distort the loss of minority classes, thus preventing meaningful representations to be learned. Motivated by this, we develop a novel joint clustering and contrastive learning framework by adapting the debiased contrastive loss to avoid under-clustering minority classes of imbalanced datasets. We show that our proposed modified debiased contrastive loss and divergence clustering loss improves the performance across multiple datasets and learning tasks. The source code is available at https://anonymous.4open.science/r/SSL-debiased-clustering

cs.CV↗