AIConfPaper
← Search

Ehsan Abbasnejad

34 accepted papers

2026

Certified but Fooled! Breaking Certified Defenses with Ghost Certificates

Quoc Viet Vo, Tashreque Mohammed Haq, Paul Montague, Tamas Abraham, Ehsan Abbasnejad, Damith C. Ranasinghe
AAAI 2026technical

Certified defenses promise provable robustness guarantees. We study the malicious exploitation of probabilistic certification frameworks to better understand the limits of guarantee provisions. Now, the objective is to not only mislead a classifier, but also to manipulate the certification process t

Cited by 0Source ↗PDF ↗Scholar ↗
2026

What Makes a Good Representation for Single-Cell Perturbation Prediction?

Wenkang Jiang, Yuhang Liu, Yichao Cai, Erdun Gao, Jiayi Dong, Ehsan Abbasnejad +2
ICML 2026poster

Single-cell perturbation modeling is fundamental for understanding and predicting cellular responses to genetic perturbations. However, existing approaches, from causal representation learning to foundation models, often struggle with an overlooked challenge: gene expression is dominated by perturba…

Cited by 0Source ↗Scholar ↗
2025

Bayesian Low-Rank Learning (Bella): A Practical Approach to Bayesian Neural Networks

Bao Gia Doan, Afshar Shamsi, Xiao-Yu Guo, Arash Mohammadi, Hamid Alinejad-Rokny, Dino Sejdinovic +3
AAAI 2025technical

Computational complexity of Bayesian learning is impeding its adoption in practical, large-scale tasks, despite demonstrations of significant merits such as improved robustness and resilience to unseen or out-of-distribution inputs over their non-Bayesian counterparts. Although, Deep ensemble method…

Cited by 5Source ↗PDF ↗code ↗Scholar ↗
2025

Do Deep Neural Network Solutions Form a Star Domain?

Ankit Sonthalia, Alexander Rubinstein, Ehsan Abbasnejad, Seong Joon Oh
ICLR 2025poster

It has recently been conjectured that neural network solution sets reachable via stochastic gradient descent (SGD) are convex, considering permutation invariances. This means that a linear path can connect two independent solutions with low loss, given the weights of one of the models are appropriat…

Cited by 1Source ↗PDF ↗code ↗Scholar ↗
2025

Do We Always Need the Simplicity Bias? Looking for Optimal Inductive Biases in the Wild

Damien Teney, Liangze Jiang, Florin Gogianu, Ehsan Abbasnejad
CVPR 2025poster

Common choices of architecture give neural networks a preference for fitting data with simple functions. This simplicity bias is known as key to their success. This paper explores the limits of this assumption. Building on recent work that showed that activation functions are the origin of the simpl…

Cited by 1Source ↗PDF ↗code ↗Scholar ↗
2025

Geometry-Aware Collaborative Multi-Solutions Optimizer for Model Fine-Tuning with Parameter Efficiency

Van-Anh Nguyen, Trung Le, Mehrtash Harandi, Ehsan Abbasnejad, Thanh-Toan Do, Dinh Phung
NeurIPS 2025poster

We propose a framework grounded in gradient flow theory and informed by geometric structure that provides multiple diverse solutions for a given task, ensuring collaborative results that enhance performance and adaptability across different tasks. This framework enables flexibility, allowing for eff…

Cited by 0Source ↗Scholar ↗
2025

Mysteries of the Deep: Role of Intermediate Representations in Out of Distribution Detection

Imezadelajara, Cristian Rodriguez-Opazo, Damien Teney, Damith Ranasinghe, Ehsan Abbasnejad
NeurIPS 2025poster

Out-of-distribution (OOD) detection is essential for reliably deploying machine learning models in the wild. Yet, most methods treat large pre-trained models as monolithic encoders and rely solely on their final-layer representations for detection. We challenge this wisdom. We reveal the intermediat…

Cited by 0Source ↗Scholar ↗
2025

RandLoRA: Full rank parameter-efficient fine-tuning of large models

Paul Albert, Frederic Z. Zhang, Hemanth Saratchandran, Cristian Rodriguez-Opazo, Anton van den Hengel, Ehsan Abbasnejad
ICLR 2025poster

Low-Rank Adaptation (LoRA) and its variants have shown impressive results in reducing the number of trainable parameters and memory requirements of large transformer networks while maintaining fine-tuning performance. The low-rank nature of the weight update inherently limits the representation powe…

Cited by 0Source ↗PDF ↗Scholar ↗
2025

Synergy and Diversity in CLIP: Enhancing Performance Through Adaptive Backbone Ensembling

Cristian Rodriguez-Opazo, Ehsan Abbasnejad, Damien Teney, Hamed Damirchi, Edison Marrese-Taylor, Anton van den Hengel
ICLR 2025poster

Contrastive Language-Image Pretraining (CLIP) stands out as a prominent method for image representation learning. Various architectures, from vision transformers~(ViTs) to convolutional networks (ResNets) have been trained with CLIP to serve as general solutions to diverse vision tasks. This paper e…

Cited by 1Source ↗PDF ↗Scholar ↗
2025

Towards Higher Effective Rank in Parameter-Efficient Fine-tuning using Khatri-Rao Product

Paul Albert, Frederic Z. Zhang, Hemanth Saratchandran, Anton van den Hengel, Ehsan Abbasnejad
ICCV 2025poster

Parameter-efficient fine-tuning (PEFT) has become a standard for adapting large pre-trained models. While low-rank adaptation (LoRA) has achieved notable success, recent studies highlight its limitations when compared to full-rank variants, particularly when scaling to demanding tasks such as vision…

Cited by 0Source ↗PDF ↗code ↗Scholar ↗
2024

BLURD: Benchmarking and Learning using a Unified Rendering and Diffusion Model

Boris Repasky, Ehsan Abbasnejad, Anthony Dick
NeurIPS 2024poster

Recent advancements in pre-trained vision models have made them pivotal in computer vision, emphasizing the need for their thorough evaluation and benchmarking. This evaluation needs to consider various factors of variation, their potential biases, shortcuts, and inaccuracies that ultimately lead to…

Cited by 0Source ↗PDF ↗code ↗Scholar ↗
2024

BRUSLEATTACK: A QUERY-EFFICIENT SCORE- BASED BLACK-BOX SPARSE ADVERSARIAL ATTACK

Quoc Viet Vo, Ehsan Abbasnejad, Damith Ranasinghe
ICLR 2024poster

We study the unique, less-well understood problem of generating sparse adversarial samples simply by observing the score-based replies to model queries. Sparse attacks aim to discover a minimum number—the $l_0$ bounded—perturbations to model inputs to craft adversarial examples and misguide model de…

Cited by 8Source ↗PDF ↗Scholar ↗
2024

Knowledge Composition using Task Vectors with Learned Anisotropic Scaling

Frederic Z. Zhang, Paul Albert, Cristian Rodriguez-Opazo, Anton van den Hengel, Ehsan Abbasnejad
NeurIPS 2024poster

Pre-trained models produce strong generic representations that can be adapted via fine-tuning on specialised datasets. The learned weight difference relative to the pre-trained model, known as a task vector, characterises the direction and stride of fine-tuning that enables the model to capture thes…

Cited by 7Source ↗PDF ↗code ↗Scholar ↗
2024

Neural Redshift: Random Networks are not Random Functions

Damien Teney, Armand Mihai Nicolicioiu, Valentin Hartmann, Ehsan Abbasnejad
CVPR 2024poster

Our understanding of the generalization capabilities of neural networks NNs is still incomplete. Prevailing explanations are based on implicit biases of gradient descent GD but they cannot account for the capabilities of models from gradientfree methods nor the simplicity bias recently observed in u…

Cited by 19Source ↗PDF ↗Scholar ↗
2024

Selective Mixup Helps with Distribution Shifts, But Not (Only) because of Mixup

Damien Teney, Jindong Wang, Ehsan Abbasnejad
ICML 2024poster

Mixup is a highly successful technique to improve generalization by augmenting training data with combinations of random pairs. Selective mixup is a family of methods that apply mixup to specific pairs e.g. combining examples across classes or domains. Despite remarkable performance on benchmarks wi…

Cited by 7Source ↗PDF ↗Scholar ↗
2024

Semantic Role Labeling Guided Out-of-distribution Detection

Jinan Zou, Maihao Guo, Yu Tian, Yuhao Lin, Haiyao Cao, Lingqiao Liu +2
COLING 2024main

Identifying unexpected domain-shifted instances in natural language processing is crucial in real-world applications. Previous works identify the out-of-distribution (OOD) instance by leveraging a single global feature embedding to represent the sentence, which cannot characterize subtle OOD pattern…

Cited by 2Source ↗PDF ↗code ↗Scholar ↗
2023

Energy-based Self-Training and Normalization for Unsupervised Domain Adaptation

Samitha Herath, Basura Fernando, Ehsan Abbasnejad, Munawar Hayat, Shahram Khadivi, Mehrtash Harandi +2
ICCV 2023poster

We propose an Unsupervised Domain Adaptation (UDA) method by making use of Energy-Based Learning (EBL) and demonstrate 1. EBL can be used to improve the instance selection for a self-training task on the unlabelled target domain, and 2. alignment and normalizing energy scores can learn domain-invari…

Cited by 9PDF ↗Scholar ↗
2023

Feature-Space Bayesian Adversarial Learning Improved Malware Detector Robustness

Bao Gia Doan, Shuiqiao Yang, Paul Montague, Olivier De Vel, Tamas Abraham, Seyit Camtepe +3
AAAI 2023technical

We present a new algorithm to train a robust malware detector. Malware is a prolific problem and malware detectors are a front-line defense. Modern detectors rely on machine learning algorithms. Now, the adversarial objective is to devise alterations to the malware code to decrease the chance of bei…

Cited by 11Source ↗PDF ↗Scholar ↗
2023

ID and OOD Performance Are Sometimes Inversely Correlated on Real-world Datasets

Damien Teney, LIN Yong, Seong Joon Oh, Ehsan Abbasnejad
NeurIPS 2023spotlight

Several studies have compared the in-distribution (ID) and out-of-distribution (OOD) performance of models in computer vision and NLP. They report a frequent positive correlation and some surprisingly never even observe an inverse correlation indicative of a necessary trade-off. The possibility of i…

Cited by 52Source ↗PDF ↗Scholar ↗
2023

ProtoCon: Pseudo-Label Refinement via Online Clustering and Prototypical Consistency for Efficient Semi-Supervised Learning

Islam Nassar, Munawar Hayat, Ehsan Abbasnejad, Hamid Rezatofighi, Gholamreza Haffari
CVPR 2023highlight

Confidence-based pseudo-labeling is among the dominant approaches in semi-supervised learning (SSL). It relies on including high-confidence predictions made on unlabeled data as additional targets to train the model. We propose ProtoCon, a novel SSL method aimed at the less-explored label-scarce SSL…

Cited by 29Source ↗PDF ↗Scholar ↗
2023

RanPAC: Random Projections and Pre-trained Models for Continual Learning

Mark McDonnell, Dong Gong, Amin Parvaneh, Ehsan Abbasnejad, Anton van den Hengel
NeurIPS 2023poster

Continual learning (CL) aims to incrementally learn different tasks (such as classification) in a non-stationary data stream without forgetting old ones. Most CL works focus on tackling catastrophic forgetting under a learning-from-scratch paradigm. However, with the increasing prominence of foundat…

Cited by 117Source ↗PDF ↗code ↗Scholar ↗
2022

Active Learning by Feature Mixing

Amin Parvaneh, Ehsan Abbasnejad, Damien Teney, Gholamreza (Reza) Haffari, Anton van den Hengel, Javen Qinfeng Shi
CVPR 2022poster

The promise of active learning (AL) is to reduce labelling costs by selecting the most valuable examples to annotate from a pool of unlabelled data. Identifying these examples is especially challenging with high-dimensional data (e. g. images, videos) and in low-data regimes. In this paper, we propo…

Cited by 132PDF ↗code ↗Scholar ↗
2022

Evading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions With Superior OOD Generalization

Damien Teney, Ehsan Abbasnejad, Simon Lucey, Anton van den Hengel
CVPR 2022poster

Neural networks trained with SGD were recently shown to rely preferentially on linearly-predictive features and can ignore complex, equally-predictive ones. This simplicity bias can explain their lack of robustness out of distribution (OOD). The more complex the task to learn, the more likely it is…

Cited by 108PDF ↗code ↗Scholar ↗
2022

Predicting Is Not Understanding: Recognizing and Addressing Underspecification in Machine Learning

Damien Teney, Maxime Peyrard, Ehsan Abbasnejad
ECCV 2022poster

"Machine learning models are typically designed for maximum accuracy on validation data. This predictive criterion rarely captures all desirable properties, in particular how a model matches a domain expert’s \emph{understanding} of the task. In this situation, known as underspecification, two model…

Cited by 33Source ↗PDF ↗Scholar ↗
2022

Progressive Class Semantic Matching for Semi-supervised Text Classification

Haiming Xu, Lingqiao Liu, Ehsan Abbasnejad
NAACL 2022long

Semi-supervised learning is a promising way to reduce the annotation cost for text-classification. Combining with pre-trained language models (PLMs), e.g., BERT, recent semi-supervised learning methods achieved impressive performance. In this work, we further investigate the marriage between semi-su…

Cited by 14Source ↗PDF ↗code ↗Scholar ↗
2021

All Labels Are Not Created Equal: Enhancing Semi-Supervision via Label Grouping and Co-Training

Islam Nassar, Samitha Herath, Ehsan Abbasnejad, Wray Buntine, Gholamreza Haffari
CVPR 2021poster

Pseudo-labeling is a key component in semi-supervised learning (SSL). It relies on iteratively using the model to generate artificial labels for the unlabeled data to train against. A common property among its various methods is that they only rely on the model's prediction to make labeling decision…

Cited by 71PDF ↗code ↗Scholar ↗
2021

Unshuffling Data for Improved Generalization in Visual Question Answering

Damien Teney, Ehsan Abbasnejad, Anton van den Hengel
ICCV 2021poster

Generalization beyond the training distribution is a core challenge in machine learning. The common practice of mixing and shuffling examples when training neural networks may not be optimal in this regard. We show that partitioning the data into well-chosen, non-i.i.d. subsets treated as multiple t…

Cited by 125PDF ↗Scholar ↗
2020

Counterfactual Vision-and-Language Navigation: Unravelling the Unseen

Amin Parvaneh, Ehsan Abbasnejad, Damien Teney, Javen Qinfeng Shi, Anton van den Hengel
NeurIPS 2020spotlight

The task of vision-and-language navigation (VLN) requires an agent to follow text instructions to find its way through simulated household environments. A prominent challenge is to train an agent capable of generalising to new environments at test time, rather than one that simply memorises trajecto…

Cited by 45Source ↗PDF ↗Scholar ↗
2020

Gold Seeker: Information Gain From Policy Distributions for Goal-Oriented Vision-and-Langauge Reasoning

Ehsan Abbasnejad, Iman Abbasnejad, Qi Wu, Javen Shi, Anton van den Hengel
CVPR 2020poster

As Computer Vision moves from passive analysis of pixels to active analysis of semantics, the breadth of information algorithms need to reason over has expanded significantly. One of the key challenges in this vein is the ability to identify the information required to make a decision, and select an…

Cited by 6PDF ↗Scholar ↗
2020

On the Value of Out-of-Distribution Testing: An Example of Goodhart's Law

Damien Teney, Ehsan Abbasnejad, Kushal Kafle, Robik Shrestha, Christopher Kanan, Anton van den Hengel
NeurIPS 2020poster

Out-of-distribution (OOD) testing is increasingly popular for evaluating a machine learning system's ability to generalize beyond the biases of a training set. OOD benchmarks are designed to present a different joint distribution of data and labels between training and test time. VQA-CP has become t…

Cited by 173Source ↗PDF ↗Scholar ↗
2019

What's to Know? Uncertainty as a Guide to Asking Goal-Oriented Questions

Ehsan Abbasnejad, Qi Wu, Qinfeng Shi, Anton van den Hengel
CVPR 2019poster

One of the core challenges in Visual Dialogue problems is asking the question that will provide the most useful information towards achieving the required objective. Encouraging an agent to ask the right questions is difficult because we don't know a-priori what information the agent will need to a…

Cited by 22PDF ↗Scholar ↗
2017

DeepSetNet: Predicting Sets With Deep Neural Networks

S. Hamid Rezatofighi, Vijay Kumar B G, Anton Milan, Ehsan Abbasnejad, Anthony Dick, Ian Reid
ICCV 2017spotlight

This paper addresses the task of set prediction using deep learning. This is important because the output of many computer vision tasks, including image tagging and object detection, are naturally expressed as sets of entities rather than vectors. As opposed to a vector, the size of a set is not fix…

Cited by 56PDF ↗Scholar ↗