Mohammadreza Ebrahimi

Mohammadreza Ebrahimi, Yidong Chai, Hao Helen Zhang et al.

Learning predictive models in new domains with scarce training data is a growing challenge in modern supervised learning scenarios. This incentivizes developing domain adaptation methods that leverage the knowledge in known domains (source) and adapt to new domains (target) with a different probability distribution. This becomes more challenging when the source and target domains are in heterogeneous feature spaces, known as heterogeneous domain adaptation (HDA). While most HDA methods utilize mathematical optimization to map source and target data to a common space, they suffer from low transferability. Neural representations have proven to be more transferable; however, they are mainly designed for homogeneous environments. Drawing on the theory of domain adaptation, we propose a novel framework, Heterogeneous Adversarial Neural Domain Adaptation (HANDA), to effectively maximize the transferability in heterogeneous environments. HANDA conducts feature and distribution alignment in a unified neural network architecture and achieves domain invariance through adversarial kernel learning. Three experiments were conducted to evaluate the performance against the state-of-the-art HDA methods on major image and text e-commerce benchmarks. HANDA shows statistically significant improvement in predictive performance. The practical utility of HANDA was shown in real-world dark web online markets. HANDA is an important step towards successful domain adaptation in e-commerce applications.

Jeremiah Birrell, Reza Ebrahimi · utoronto

We introduce a new class of optimal-transport-regularized divergences, $D^c$, constructed via an infimal convolution between an information divergence, $D$, and an optimal-transport (OT) cost, $C$, and study their use in distributionally robust optimization (DRO). In particular, we propose the $ARMOR_D$ methods as novel approaches to enhancing the adversarial robustness of deep learning models. These DRO-based methods are defined by minimizing the maximum expected loss over a $D^c$-neighborhood of the empirical distribution of the training data. Viewed as a tool for constructing adversarial samples, our method allows samples to be both transported, according to the OT cost, and re-weighted, according to the information divergence; the addition of a principled and dynamical adversarial re-weighting on top of adversarial sample transport is a key innovation of $ARMOR_D$. $ARMOR_D$ can be viewed as a generalization of the best-performing loss functions and OT costs in the adversarial training literature; we demonstrate this flexibility by using $ARMOR_D$ to augment the UDR, TRADES, and MART methods and obtain improved performance on CIFAR-10 and CIFAR-100 image recognition. Specifically, augmenting with $ARMOR_D$ leads to 1.9\% and 2.1\% improvement against AutoAttack, a powerful ensemble of adversarial attacks, on CIFAR-10 and CIFAR-100 respectively. To foster reproducibility, we made the code accessible at https://github.com/star-ailab/ARMOR.

M. Nikhil Krishnan, MohammadReza Ebrahimi, Ashish Khisti · utoronto

In distributed computing, slower nodes (stragglers) usually become a bottleneck. Gradient Coding (GC), introduced by Tandon et al., is an efficient technique that uses principles of error-correcting codes to distribute gradient computation in the presence of stragglers. In this paper, we consider the distributed computation of a sequence of gradients $\{g(1),g(2),\ldots,g(J)\}$, where processing of each gradient $g(t)$ starts in round-$t$ and finishes by round-$(t+T)$. Here $T\geq 0$ denotes a delay parameter. For the GC scheme, coding is only across computing nodes and this results in a solution where $T=0$. On the other hand, having $T>0$ allows for designing schemes which exploit the temporal dimension as well. In this work, we propose two schemes that demonstrate improved performance compared to GC. Our first scheme combines GC with selective repetition of previously unfinished tasks and achieves improved straggler mitigation. In our second scheme, which constitutes our main contribution, we apply GC to a subset of the tasks and repetition for the remainder of the tasks. We then multiplex these two classes of tasks across workers and rounds in an adaptive manner, based on past straggler patterns. Using theoretical analysis, we demonstrate that our second scheme achieves significant reduction in the computational load. In our experiments, we study a practical setting of concurrently training multiple neural networks over an AWS Lambda cluster involving 256 worker nodes, where our framework naturally applies. We demonstrate that the latter scheme can yield a 16\% improvement in runtime over the baseline GC scheme, in the presence of naturally occurring, non-simulated stragglers.

MohammadReza Ebrahimi, Sunny Panchal, Roland Memisevic

Despite their recent successes, Transformer-based large language models show surprising failure modes. A well-known example of such failure modes is their inability to length-generalize: solving problem instances at inference time that are longer than those seen during training. In this work, we further explore the root cause of this failure by performing a detailed analysis of model behaviors on the simple parity task. Our analysis suggests that length generalization failures are intricately related to a model's inability to perform random memory accesses within its context window. We present supporting evidence for this hypothesis by demonstrating the effectiveness of methodologies that circumvent the need for indexing or that enable random token access indirectly, through content-based addressing. We further show where and how the failure to perform random memory access manifests through attention map visualizations.

M. Reza Ebrahimi, Jun Chen, Ashish Khisti

This paper investigates a novel lossy compression framework operating under logarithmic loss, designed to handle situations where the reconstruction distribution diverges from the source distribution. This framework is especially relevant for applications that require joint compression and retrieval, and in scenarios involving distributional shifts due to processing. We show that the proposed formulation extends the classical minimum entropy coupling framework by integrating a bottleneck, allowing for a controlled degree of stochasticity in the coupling. We explore the decomposition of the Minimum Entropy Coupling with Bottleneck (MEC-B) into two distinct optimization problems: Entropy-Bounded Information Maximization (EBIM) for the encoder, and Minimum Entropy Coupling (MEC) for the decoder. Through extensive analysis, we provide a greedy algorithm for EBIM with guaranteed performance, and characterize the optimal solution near functional mappings, yielding significant theoretical insights into the structural complexity of this problem. Furthermore, we illustrate the practical application of MEC-B through experiments in Markov Coding Games (MCGs) under rate limits. These games simulate a communication scenario within a Markov Decision Process, where an agent must transmit a compressed message from a sender to a receiver through its actions. Our experiments highlight the trade-offs between MDP rewards and receiver accuracy across various compression rates, showcasing the efficacy of our method compared to conventional compression baseline.

Ashish Khisti, M. Reza Ebrahimi, Hassan Dbouk et al.

We consider multi-draft speculative sampling, where the proposal sequences are sampled independently from different draft models. At each step, a token-level draft selection scheme takes a list of valid tokens as input and produces an output token whose distribution matches that of the target model. Previous works have demonstrated that the optimal scheme (which maximizes the probability of accepting one of the input tokens) can be cast as a solution to a linear program. In this work we show that the optimal scheme can be decomposed into a two-step solution: in the first step an importance sampling (IS) type scheme is used to select one intermediate token; in the second step (single-draft) speculative sampling is applied to generate the output token. For the case of two identical draft models we further 1) establish a necessary and sufficient condition on the distributions of the target and draft models for the acceptance probability to equal one and 2) provide an explicit expression for the optimal acceptance probability. Our theoretical analysis also motives a new class of token-level selection schemes based on weighted importance sampling. Our experimental results demonstrate consistent improvements in the achievable block efficiency and token rates over baseline schemes in a number of scenarios.

M. Reza Ebrahimi, Roland Memisevic

The role of hidden units in recurrent neural networks is typically seen as modeling memory, with research focusing on enhancing information retention through gating mechanisms. A less explored perspective views hidden units as active participants in the computation performed by the network, rather than passive memory stores. In this work, we revisit bilinear operations, which involve multiplicative interactions between hidden units and input embeddings. We demonstrate theoretically and empirically that they constitute a natural inductive bias for representing the evolution of hidden states in state tracking tasks. These are the simplest type of tasks that require hidden units to actively contribute to the behavior of the network. We also show that bilinear state updates form a natural hierarchy corresponding to state tracking tasks of increasing complexity, with popular linear recurrent networks such as Mamba residing at the lowest-complexity center of that hierarchy.

Buu Phan, Reza Ebrahimi, Sanjay Haresh et al.

We study length generalization in sequence models on a composite problem involving both state tracking and associative recall. Prior work finds that recurrent networks handle state tracking well but struggle with recall, whereas Transformers excel at recall yet fail to extend state-tracking capabilities to longer sequences. Motivated by the complementary strengths of these architectures, we construct hybrid models integrating recurrent and attention-based components, and train them on the combined task to evaluate whether both capabilities can be preserved. Our results reveal that, in such hybrids, the Transformer component tends to exploit shortcut solutions, leading to poor length generalization. We identify this shortcut reliance as a key obstacle and propose a simple yet effective training strategy -- delaying the training of the attention layers -- that mitigates this effect and significantly improves length generalization performance. Our experiments show that this approach enables hybrid models to achieve near-perfect accuracy ($>90\%$) on hybrid sequences three times longer than those used during training.

Yizhi Liu, Fang Yu Lin, Mohammadreza Ebrahimi et al.

Internet users have been exposing an increasing amount of Personally Identifiable Information (PII) on social media. Such exposed PII can cause severe losses to the users, and informing users of their PII exposure is crucial to raise their privacy awareness and encourage them to take protective measures. To this end, advanced automatic techniques are needed. While Information Extraction (IE) techniques can be used to extract the PII automatically, Deep Learning (DL)-based IE models alleviate the need for feature engineering and further improve the efficiency. However, DL-based IE models often require large-scale labeled data for training, but PII-labeled social media posts are difficult to obtain due to privacy concerns. Also, these models rely heavily on pre-trained word embeddings, while PII in social media often varies in forms and thus has no fixed representations in pre-trained word embeddings. In this study, we propose the Deep Transfer Learning for PII Extraction (DTL-PIIE) framework to address these two limitations. DTL-PIIE transfers knowledge learned from publicly available PII data to social media to address the problem of rare PII-labeled data. Moreover, our framework leverages Graph Convolutional Networks (GCNs) to incorporate syntactic patterns to guide PIIE without relying on pre-trained word embeddings. Evaluation against benchmark IE models indicates that our approach outperforms state-of-the-art DL-based IE models. Our framework can facilitate various applications, such as PII misuse prediction and privacy risk assessment, protecting the privacy of internet users.

Mohammadreza Ebrahimi, Ning Zhang, James Hu et al.

Anti-malware engines are the first line of defense against malicious software. While widely used, feature engineering-based anti-malware engines are vulnerable to unseen (zero-day) attacks. Recently, deep learning-based static anti-malware detectors have achieved success in identifying unseen attacks without requiring feature engineering and dynamic analysis. However, these detectors are susceptible to malware variants with slight perturbations, known as adversarial examples. Generating effective adversarial examples is useful to reveal the vulnerabilities of such systems. Current methods for launching such attacks require accessing either the specifications of the targeted anti-malware model, the confidence score of the anti-malware response, or dynamic malware analysis, which are either unrealistic or expensive. We propose MalRNN, a novel deep learning-based approach to automatically generate evasive malware variants without any of these restrictions. Our approach features an adversarial example generation process, which learns a language model via a generative sequence-to-sequence recurrent neural network to augment malware binaries. MalRNN effectively evades three recent deep learning-based malware detectors and outperforms current benchmark methods. Findings from applying our MalRNN on a real dataset with eight malware categories are discussed.

MohammadReza Ebrahimi, Navona Calarco, Kieran Campbell et al.

Multi-subject fMRI studies are challenging due to the high variability of both brain anatomy and functional brain topographies across participants. An effective way of aggregating multi-subject fMRI data is to extract a shared representation that filters out unwanted variability among subjects. Some recent work has implemented probabilistic models to extract a shared representation in task fMRI. In the present work, we improve upon these models by incorporating temporal information in the common latent structures. We introduce a new model, Shared Gaussian Process Factor Analysis (S-GPFA), that discovers shared latent trajectories and subject-specific functional topographies, while modelling temporal correlation in fMRI data. We demonstrate the efficacy of our model in revealing ground truth latent structures using simulated data, and replicate experimental performance of time-segment matching and inter-subject similarity on the publicly available Raider and Sherlock datasets. We further test the utility of our model by analyzing its learned model parameters in the large multi-site SPINS dataset, on a social cognition task from participants with and without schizophrenia.