arXiv subjects

Shayan Mohajer Hamidi

Publications and source records attributed to Shayan Mohajer Hamidi.

At least 19 recordsLinked to original sources

arXiv2026

Smaller Models, Better Rejects: Preference Distillation Scaling

Preference distillation typically treats a teacher response as preferred and the student's own response as rejected. This assumes that self-generated failures are the most informative negatives and that rejects must come from a model at least as large as the student, making generation costly at scale. We find neither assumption holds: across students from 7B to 72B, smaller frozen models generate rejects with less inference compute yet train stronger students than self-generated rejects, before and after sequence-level knowledge distillation, on code generation and mathematical reasoning. To explain this result, we derive a finite-horizon utility bound for Direct Preference Optimization in a linearized feature model. The bound characterizes favorable reject distributions and motivates three interventions. First, mixing rejects from smaller and student-scale models improves performance as the smaller model's share increases. Second, reassigning rejects to other prompts and shuffling their code tokens still outperform length-matched gibberish, showing that task structure contributes to reject utility. Third, selecting candidates with lower likelihood under the reference policy improves net transfer when higher-likelihood candidates provide less useful contrast. Lower-likelihood selections outperform higher-likelihood ones for every source. These results suggest that effective rejects preserve task structure while limiting coupling to the reference policy, and that smaller frozen models can provide them at low cost.

cs.LG↗

arXiv2026

CARA: Collision-Aware Resolution Adaptation for Multiresolution Hash Encoding Based Image Fitting

Multiresolution hash encodings have recently enabled fast and high-fidelity implicit neural representations by storing multi-scale features in fixed-size hash tables along a geometric resolution schedule. However, the standard design is data-agnostic: different resolution levels receive identical hash-table capacity despite large differences in image frequency content. As a result, some levels experience severe hash collisions while others underutilize parameters, leading to inefficient capacity allocation. To address this issue, we propose Collision-Aware Resolution Adaptation (CARA), a method that assigns per-level resolutions by balancing the effective information load across hash levels. This adaptive allocation reduces capacity bottlenecks and improves parameter efficiency. In addition, we introduce an invertible pixel-shuffle transform that reduces hash load factors by redistributing spatial information, thereby mitigating collision-induced information loss without enlarging the hash tables. To support evaluation on extremely high-resolution data, we also curate, to the best of our knowledge, the first uncompressed whole-slide image dataset for academic research. Experiments on Kodak images, gigapixel natural images, and raw whole-slide images demonstrate that CARA consistently improves the fidelity-parameter trade-off. Our method matches state-of-the-art performance while using only $27.76%$ of the parameters, and achieves up to $6.11$ dB PSNR improvement at comparable parameter counts. Code is provided in the supplementary.

cs.CV↗

arXiv2026

Regime-aware financial volatility forecasting via in-context learning

This work introduces a regime-aware in-context learning framework that leverages large language models (LLMs) for financial volatility forecasting under nonstationary market conditions. The proposed approach deploys pretrained LLMs to reason over historical volatility patterns and adjust their predictions without parameter fine-tuning. We develop an oracle-guided refinement procedure that constructs regime-aware demonstrations from training data. An LLM is then deployed as an in-context learner that predicts the next-step volatility from the input sequence using demonstrations sampled conditional to the estimated market label. This conditional sampling strategy enables the LLM to adapt its predictions to regime-dependent volatility dynamics through contextual reasoning alone. Experiments with multiple financial datasets show that the proposed regime-aware in-context learning framework outperforms both classical volatility forecasting approaches and direct one-shot learning, especially during high-volatility periods.

cs.LG↗

arXiv2026

ASMIL: Attention-Stabilized Multiple Instance Learning for Whole Slide Imaging

Attention-based multiple instance learning (MIL) has emerged as a powerful framework for whole slide image (WSI) diagnosis, leveraging attention to aggregate instance-level features into bag-level predictions. Despite this success, we find that such methods exhibit a new failure mode: unstable attention dynamics. Across four representative attention-based MIL methods and two public WSI datasets, we observe that attention distributions oscillate across epochs rather than converging to a consistent pattern, degrading performance. This instability adds to two previously reported challenges: overfitting and over-concentrated attention distribution. To simultaneously overcome these three limitations, we introduce attention-stabilized multiple instance learning (ASMIL), a novel unified framework. ASMIL uses an anchor model to stabilize attention, replaces softmax with a normalized sigmoid function in the anchor to prevent over-concentration, and applies token random dropping to mitigate overfitting. Extensive experiments demonstrate that ASMIL achieves up to a 6.49\% F1 score improvement over state-of-the-art methods. Moreover, integrating the anchor model and normalized sigmoid into existing attention-based MIL methods consistently boosts their performance, with F1 score gains up to 10.73\%. All code and data are publicly available at https://github.com/Linfeng-Ye/ASMIL.

cs.CV↗

arXiv2026

Coupled Data and Measurement Space Dynamics for Enhanced Diffusion Posterior Sampling

Inverse problems, where the goal is to recover an unknown signal from noisy or incomplete measurements, are central to applications in medical imaging, remote sensing, and computational biology. Diffusion models have recently emerged as powerful priors for solving such problems. However, existing methods either rely on projection-based techniques that enforce measurement consistency through heuristic updates, or they approximate the likelihood $p(\boldsymbol{y} \mid \boldsymbol{x})$, often resulting in artifacts and instability under complex or high-noise conditions. To address these limitations, we propose a novel framework called \emph{coupled data and measurement space diffusion posterior sampling} (C-DPS), which eliminates the need for constraint tuning or likelihood approximation. C-DPS introduces a forward stochastic process in the measurement space $\{\boldsymbol{y}_t\}$, evolving in parallel with the data-space diffusion $\{\boldsymbol{x}_t\}$, which enables the derivation of a closed-form posterior $p(\boldsymbol{x}_{t-1} \mid \boldsymbol{x}_t, \boldsymbol{y}_{t-1})$. This coupling allows for accurate and recursive sampling based on a well-defined posterior distribution. Empirical results demonstrate that C-DPS consistently outperforms existing baselines, both qualitatively and quantitatively, across multiple inverse problem benchmarks.

cs.LG↗

arXiv2025

Information-Guided Diffusion Sampling for Dataset Distillation

Dataset distillation aims to create a compact dataset that retains essential information while maintaining model performance. Diffusion models (DMs) have shown promise for this task but struggle in low images-per-class (IPC) settings, where generated samples lack diversity. In this paper, we address this issue from an information-theoretic perspective by identifying two key types of information that a distilled dataset must preserve: ($i$) prototype information $\mathrm{I}(X;Y)$, which captures label-relevant features; and ($ii$) contextual information $\mathrm{H}(X | Y)$, which preserves intra-class variability. Here, $(X,Y)$ represents the pair of random variables corresponding to the input data and its ground truth label, respectively. Observing that the required contextual information scales with IPC, we propose maximizing $\mathrm{I}(X;Y) + β\mathrm{H}(X | Y)$ during the DM sampling process, where $β$ is IPC-dependent. Since directly computing $\mathrm{I}(X;Y)$ and $\mathrm{H}(X | Y)$ is intractable, we develop variational estimations to tightly lower-bound these quantities via a data-driven approach. Our approach, information-guided diffusion sampling (IGDS), seamlessly integrates with diffusion models and improves dataset distillation across all IPC settings. Experiments on Tiny ImageNet and ImageNet subsets show that IGDS significantly outperforms existing methods, particularly in low-IPC regimes. The code will be released upon acceptance.

cs.LG↗

arXiv2025

Towards Undistillable Models by Minimizing Conditional Mutual Information

A deep neural network (DNN) is said to be undistillable if, when used as a black-box input-output teacher, it cannot be distilled through knowledge distillation (KD). In this case, the distilled student (referred to as the knockoff student) does not outperform a student trained independently with label smoothing (LS student) in terms of prediction accuracy. To protect intellectual property of DNNs, it is desirable to build undistillable DNNs. To this end, it is first observed that an undistillable DNN may have the trait that each cluster of its output probability distributions in response to all sample instances with the same label should be highly concentrated to the extent that each cluster corresponding to each label should ideally collapse into one probability distribution. Based on this observation and by measuring the concentration of each cluster in terms of conditional mutual information (CMI), a new training method called CMI minimized (CMIM) method is proposed, which trains a DNN by jointly minimizing the conventional cross entropy (CE) loss and the CMI values of all temperature scaled clusters across the entire temperature spectrum. The resulting CMIM model is shown, by extensive experiments, to be undistillable by all tested KD methods existing in the literature. That is, the knockoff students distilled by these KD methods from the CMIM model underperform the respective LS students. In addition, the CMIM model is also shown to performs better than the model trained with the CE loss alone in terms of their own prediction accuracy.

cs.LG↗

arXiv2025

Distributed Quasi-Newton Method for Fair and Fast Federated Learning

Federated learning (FL) is a promising technology that enables edge devices/clients to collaboratively and iteratively train a machine learning model under the coordination of a central server. The most common approach to FL is first-order methods, where clients send their local gradients to the server in each iteration. However, these methods often suffer from slow convergence rates. As a remedy, second-order methods, such as quasi-Newton, can be employed in FL to accelerate its convergence. Unfortunately, similarly to the first-order FL methods, the application of second-order methods in FL can lead to unfair models, achieving high average accuracy while performing poorly on certain clients' local datasets. To tackle this issue, in this paper we introduce a novel second-order FL framework, dubbed \textbf{d}istributed \textbf{q}uasi-\textbf{N}ewton \textbf{fed}erated learning (DQN-Fed). This approach seeks to ensure fairness while leveraging the fast convergence properties of quasi-Newton methods in the FL context. Specifically, DQN-Fed helps the server update the global model in such a way that (i) all local loss functions decrease to promote fairness, and (ii) the rate of change in local loss functions aligns with that of the quasi-Newton method. We prove the convergence of DQN-Fed and demonstrate its \textit{linear-quadratic} convergence rate. Moreover, we validate the efficacy of DQN-Fed across a range of federated datasets, showing that it surpasses state-of-the-art fair FL methods in fairness, average accuracy and convergence speed.

cs.LG↗

arXiv2025

Coded Deep Learning: Framework and Algorithm

The success of deep learning (DL) is often achieved with large models and high complexity during both training and post-training inferences, hindering training in resource-limited settings. To alleviate these issues, this paper introduces a new framework dubbed ``coded deep learning'' (CDL), which integrates information-theoretic coding concepts into the inner workings of DL, to significantly compress model weights and activations, reduce computational complexity at both training and post-training inference stages, and enable efficient model/data parallelism. Specifically, within CDL, (i) we first propose a novel probabilistic method for quantizing both model weights and activations, and its soft differentiable variant which offers an analytic formula for gradient calculation during training; (ii) both the forward and backward passes during training are executed over quantized weights and activations, eliminating most floating-point operations and reducing training complexity; (iii) during training, both weights and activations are entropy constrained so that they are compressible in an information-theoretic sense throughout training, thus reducing communication costs in model/data parallelism; and (iv) the trained model in CDL is by default in a quantized format with compressible quantized weights, reducing post-training inference and storage complexity. Additionally, a variant of CDL, namely relaxed CDL (R-CDL), is presented to further improve the trade-off between validation accuracy and compression though requiring full precision in training with other advantageous features of CDL intact. Extensive empirical results show that CDL and R-CDL outperform the state-of-the-art algorithms in DNN compression in the literature.

cs.LG↗

arXiv2025

Conditional Mutual Information Based Diffusion Posterior Sampling for Solving Inverse Problems

Inverse problems are prevalent across various disciplines in science and engineering. In the field of computer vision, tasks such as inpainting, deblurring, and super-resolution are commonly formulated as inverse problems. Recently, diffusion models (DMs) have emerged as a promising approach for addressing noisy linear inverse problems, offering effective solutions without requiring additional task-specific training. Specifically, with the prior provided by DMs, one can sample from the posterior by finding the likelihood. Since the likelihood is intractable, it is often approximated in the literature. However, this approximation compromises the quality of the generated images. To overcome this limitation and improve the effectiveness of DMs in solving inverse problems, we propose an information-theoretic approach. Specifically, we maximize the conditional mutual information $\mathrm{I}(\boldsymbol{x}_0; \boldsymbol{y} | \boldsymbol{x}_t)$, where $\boldsymbol{x}_0$ represents the reconstructed signal, $\boldsymbol{y}$ is the measurement, and $\boldsymbol{x}_t$ is the intermediate signal at stage $t$. This ensures that the intermediate signals $\boldsymbol{x}_t$ are generated in a way that the final reconstructed signal $\boldsymbol{x}_0$ retains as much information as possible about the measurement $\boldsymbol{y}$. We demonstrate that this method can be seamlessly integrated with recent approaches and, once incorporated, enhances their performance both qualitatively and quantitatively.

cs.LG↗

arXiv2025

Over-the-Air Fair Federated Learning via Multi-Objective Optimization

In federated learning (FL), heterogeneity among the local dataset distributions of clients can result in unsatisfactory performance for some, leading to an unfair model. To address this challenge, we propose an over-the-air fair federated learning algorithm (OTA-FFL), which leverages over-the-air computation to train fair FL models. By formulating FL as a multi-objective minimization problem, we introduce a modified Chebyshev approach to compute adaptive weighting coefficients for gradient aggregation in each communication round. To enable efficient aggregation over the multiple access channel, we derive analytical solutions for the optimal transmit scalars at the clients and the de-noising scalar at the parameter server. Extensive experiments demonstrate the superiority of OTA-FFL in achieving fairness and robust performance compared to existing methods.

cs.LG↗

arXiv2024

Enhancing Diffusion Models for Inverse Problems with Covariance-Aware Posterior Sampling

Inverse problems exist in many disciplines of science and engineering. In computer vision, for example, tasks such as inpainting, deblurring, and super resolution can be effectively modeled as inverse problems. Recently, denoising diffusion probabilistic models (DDPMs) are shown to provide a promising solution to noisy linear inverse problems without the need for additional task specific training. Specifically, with the prior provided by DDPMs, one can sample from the posterior by approximating the likelihood. In the literature, approximations of the likelihood are often based on the mean of conditional densities of the reverse process, which can be obtained using Tweedie formula. To obtain a better approximation to the likelihood, in this paper we first derive a closed form formula for the covariance of the reverse process. Then, we propose a method based on finite difference method to approximate this covariance such that it can be readily obtained from the existing pretrained DDPMs, thereby not increasing the complexity compared to existing approaches. Finally, based on the mean and approximated covariance of the reverse process, we present a new approximation to the likelihood. We refer to this method as covariance-aware diffusion posterior sampling (CA-DPS). Experimental results show that CA-DPS significantly improves reconstruction performance without requiring hyperparameter tuning. The code for the paper is put in the supplementary materials.

cs.CV↗

arXiv2024

GP-FL: Model-Based Hessian Estimation for Second-Order Over-the-Air Federated Learning

Second-order methods are widely adopted to improve the convergence rate of learning algorithms. In federated learning (FL), these methods require the clients to share their local Hessian matrices with the parameter server (PS), which comes at a prohibitive communication cost. A classical solution to this issue is to approximate the global Hessian matrix from the first-order information. Unlike in idealized networks, this solution does not perform effectively in over-the-air FL settings, where the PS receives noisy versions of the local gradients. This paper introduces a novel second-order FL framework tailored for wireless channels. The pivotal innovation lies in the PS's capability to directly estimate the global Hessian matrix from the received noisy local gradients via a non-parametric method: the PS models the unknown Hessian matrix as a Gaussian process, and then uses the temporal relation between the gradients and Hessian along with the channel model to find a stochastic estimator for the global Hessian matrix. We refer to this method as Gaussian process-based Hessian modeling for wireless FL (GP-FL) and show that it exhibits a linear-quadratic convergence rate. Numerical experiments on various datasets demonstrate that GP-FL outperforms all classical baseline first and second order FL approaches.

cs.LG↗

arXiv2024

Rate-Constrained Quantization for Communication-Efficient Federated Learning

Quantization is a common approach to mitigate the communication cost of federated learning (FL). In practice, the quantized local parameters are further encoded via an entropy coding technique, such as Huffman coding, for efficient data compression. In this case, the exact communication overhead is determined by the bit rate of the encoded gradients. Recognizing this fact, this work deviates from the existing approaches in the literature and develops a novel quantized FL framework, called \textbf{r}ate-\textbf{c}onstrained \textbf{fed}erated learning (RC-FED), in which the gradients are quantized subject to both fidelity and data rate constraints. We formulate this scheme, as a joint optimization in which the quantization distortion is minimized while the rate of encoded gradients is kept below a target threshold. This enables for a tunable trade-off between quantization distortion and communication cost. We analyze the convergence behavior of RC-FED, and show its superior performance against baseline quantized FL schemes on several datasets.

cs.LG↗

arXiv2024

How to Train the Teacher Model for Effective Knowledge Distillation

Recently, it was shown that the role of the teacher in knowledge distillation (KD) is to provide the student with an estimate of the true Bayes conditional probability density (BCPD). Notably, the new findings propose that the student's error rate can be upper-bounded by the mean squared error (MSE) between the teacher's output and BCPD. Consequently, to enhance KD efficacy, the teacher should be trained such that its output is close to BCPD in MSE sense. This paper elucidates that training the teacher model with MSE loss equates to minimizing the MSE between its output and BCPD, aligning with its core responsibility of providing the student with a BCPD estimate closely resembling it in MSE terms. In this respect, through a comprehensive set of experiments, we demonstrate that substituting the conventional teacher trained with cross-entropy loss with one trained using MSE loss in state-of-the-art KD methods consistently boosts the student's accuracy, resulting in improvements of up to 2.6\%.

cs.LG↗

arXiv2024

Adversarial Training via Adaptive Knowledge Amalgamation of an Ensemble of Teachers

Adversarial training (AT) is a popular method for training robust deep neural networks (DNNs) against adversarial attacks. Yet, AT suffers from two shortcomings: (i) the robustness of DNNs trained by AT is highly intertwined with the size of the DNNs, posing challenges in achieving robustness in smaller models; and (ii) the adversarial samples employed during the AT process exhibit poor generalization, leaving DNNs vulnerable to unforeseen attack types. To address these dual challenges, this paper introduces adversarial training via adaptive knowledge amalgamation of an ensemble of teachers (AT-AKA). In particular, we generate a diverse set of adversarial samples as the inputs to an ensemble of teachers; and then, we adaptively amalgamate the logtis of these teachers to train a generalized-robust student. Through comprehensive experiments, we illustrate the superior efficacy of AT-AKA over existing AT methods and adversarial robustness distillation techniques against cutting-edge attacks, including AutoAttack.

cs.LG↗

arXiv2024

Bayes Conditional Distribution Estimation for Knowledge Distillation Based on Conditional Mutual Information

It is believed that in knowledge distillation (KD), the role of the teacher is to provide an estimate for the unknown Bayes conditional probability distribution (BCPD) to be used in the student training process. Conventionally, this estimate is obtained by training the teacher using maximum log-likelihood (MLL) method. To improve this estimate for KD, in this paper we introduce the concept of conditional mutual information (CMI) into the estimation of BCPD and propose a novel estimator called the maximum CMI (MCMI) method. Specifically, in MCMI estimation, both the log-likelihood and CMI of the teacher are simultaneously maximized when the teacher is trained. Through Eigen-CAM, it is further shown that maximizing the teacher's CMI value allows the teacher to capture more contextual information in an image cluster. Via conducting a thorough set of experiments, we show that by employing a teacher trained via MCMI estimation rather than one trained via MLL estimation in various state-of-the-art KD frameworks, the student's classification accuracy consistently increases, with the gain of up to 3.32\%. This suggests that the teacher's BCPD estimate provided by MCMI method is more accurate than that provided by MLL method. In addition, we show that such improvements in the student's accuracy are more drastic in zero-shot and few-shot settings. Notably, the student's accuracy increases with the gain of up to 5.72\% when 5\% of the training samples are available to the student (few-shot), and increases from 0\% to as high as 84\% for an omitted class (zero-shot). The code is available at \url{https://github.com/iclr2024mcmi/ICLRMCMI}.

cs.LG↗

arXiv2024

Thundernna: a white box adversarial attack

The existing work shows that the neural network trained by naive gradient-based optimization method is prone to adversarial attacks, adds small malicious on the ordinary input is enough to make the neural network wrong. At the same time, the attack against a neural network is the key to improving its robustness. The training against adversarial examples can make neural networks resist some kinds of adversarial attacks. At the same time, the adversarial attack against a neural network can also reveal some characteristics of the neural network, a complex high-dimensional non-linear function, as discussed in previous work. In This project, we develop a first-order method to attack the neural network. Compare with other first-order attacks, our method has a much higher success rate. Furthermore, it is much faster than second-order attacks and multi-steps first-order attacks.

cs.LG↗