no code implementations • NAACL (maiworkshop) 2021 • Woojeong Jin, Maziar Sanjabi, Shaoliang Nie, Liang Tan, Xiang Ren, Hamed Firooz
In this paper, we propose modality-specific distillation (MSD) to effectively transfer knowledge from a teacher on multimodal datasets.
no code implementations • 21 Mar 2024 • Jonathan Lebensold, Maziar Sanjabi, Pietro Astolfi, Adriana Romero-Soriano, Kamalika Chaudhuri, Mike Rabbat, Chuan Guo
Text-to-image diffusion models have been shown to suffer from sample-level memorization, possibly reproducing near-perfect replica of images that they are trained on, which may be undesirable.
no code implementations • 4 Mar 2024 • Tom Sander, Yaodong Yu, Maziar Sanjabi, Alain Durmus, Yi Ma, Kamalika Chaudhuri, Chuan Guo
In this work, we show that effective DP representation learning can be done via image captioning and scaling up to internet-scale multimodal datasets.
no code implementations • 17 Nov 2023 • Animesh Sinha, Bo Sun, Anmol Kalia, Arantxa Casanova, Elliot Blanchard, David Yan, Winnie Zhang, Tony Nelli, Jiahui Chen, Hardik Shah, Licheng Yu, Mitesh Kumar Singh, Ankit Ramchandani, Maziar Sanjabi, Sonal Gupta, Amy Bearman, Dhruv Mahajan
Evaluation results show our method improves visual quality by 14%, prompt alignment by 16. 2% and scene diversity by 15. 3%, compared to prompt engineering the base Emu model for stickers generation.
no code implementations • 7 Oct 2023 • Song Jiang, Zahra Shakeri, Aaron Chan, Maziar Sanjabi, Hamed Firooz, Yinglong Xia, Bugra Akyildiz, Yizhou Sun, Jinchao Li, Qifan Wang, Asli Celikyilmaz
Breakdown analysis further highlights RESPROMPT particularly excels in complex multi-step reasoning: for questions demanding at least five reasoning steps, RESPROMPT outperforms the best CoT based benchmarks by a remarkable average improvement of 21. 1% on LLaMA-65B and 14. 3% on LLaMA2-70B.
no code implementations • 3 Oct 2023 • Samyadeep Basu, Mehrdad Saberi, Shweta Bhardwaj, Atoosa Malemir Chegini, Daniela Massiceti, Maziar Sanjabi, Shell Xu Hu, Soheil Feizi
From both the human study and automated evaluation, we find that: (i) Instruct-Pix2Pix, Null-Text and SINE are the top-performing methods averaged across different edit types, however {\it only} Instruct-Pix2Pix and Null-Text are able to preserve original image properties; (ii) Most of the editing methods fail at edits involving spatial operations (e. g., changing the position of an object).
1 code implementation • 20 Jul 2023 • Neha Kalibhat, Shweta Bhardwaj, Bayan Bruss, Hamed Firooz, Maziar Sanjabi, Soheil Feizi
Although many existing approaches interpret features independently, we observe in state-of-the-art self-supervised and supervised models, that less than 20% of the representation space can be explained by individual features.
no code implementations • 18 Jul 2023 • Samyadeep Basu, Maziar Sanjabi, Daniela Massiceti, Shell Xu Hu, Soheil Feizi
On the challenging Winoground compositional reasoning benchmark, our method improves the absolute visio-linguistic performance of different CLIP models by up to 7%, while on the ARO dataset, our method improves the visio-linguistic performance by upto 3%.
1 code implementation • 15 Jun 2023 • Yaodong Yu, Maziar Sanjabi, Yi Ma, Kamalika Chaudhuri, Chuan Guo
In this work, we propose as a mitigation measure a recipe to train foundation vision models with differential privacy (DP) guarantee.
1 code implementation • 10 May 2023 • Mazda Moayeri, Keivan Rezaei, Maziar Sanjabi, Soheil Feizi
We observe that the mapping between an image's representation in one model to its representation in another can be learned surprisingly well with just a linear layer, even across diverse models.
1 code implementation • CVPR 2023 • Ajinkya Tejankar, Maziar Sanjabi, Qifan Wang, Sinong Wang, Hamed Firooz, Hamed Pirsiavash, Liang Tan
It was shown that an adversary can poison a small part of the unlabeled data so that when a victim trains an SSL model on it, the final model will have a backdoor that the adversary can exploit.
no code implementations • 24 Oct 2022 • Chuan Guo, Alexandre Sablayrolles, Maziar Sanjabi
Differential privacy (DP) is by far the most widely accepted framework for mitigating privacy risks in machine learning.
1 code implementation • 14 Oct 2022 • John Nguyen, Jianyu Wang, Kshitiz Malik, Maziar Sanjabi, Michael Rabbat
Surprisingly, we also find that starting federated learning from a pre-trained initialization reduces the effect of both data and system heterogeneity.
no code implementations • 14 Oct 2022 • Nan Wang, Qifan Wang, Yi-Chia Wang, Maziar Sanjabi, Jingzhou Liu, Hamed Firooz, Hongning Wang, Shaoliang Nie
However, the bias inherent in user written text, often used for PTG model training, can inadvertently associate different levels of linguistic quality with users' protected attributes.
no code implementations • 2 Jul 2022 • Aaron Chan, Shaoliang Nie, Liang Tan, Xiaochang Peng, Hamed Firooz, Maziar Sanjabi, Xiang Ren
Following how humans communicate, free-text rationales aim to use natural language to explain neural language model (LM) behavior.
2 code implementations • 30 Jun 2022 • John Nguyen, Jianyu Wang, Kshitiz Malik, Maziar Sanjabi, Michael Rabbat
Surprisingly, we also find that starting federated learning from a pre-trained initialization reduces the effect of both data and system heterogeneity.
1 code implementation • 25 May 2022 • Brihi Joshi, Aaron Chan, Ziyi Liu, Shaoliang Nie, Maziar Sanjabi, Hamed Firooz, Xiang Ren
to align with human rationales (Which input tokens would humans focus on?).
no code implementations • 27 Apr 2022 • Samuel Horváth, Maziar Sanjabi, Lin Xiao, Peter Richtárik, Michael Rabbat
The practice of applying several local updates before aggregation across clients has been empirically shown to be a successful approach to overcoming the communication bottleneck in Federated Learning (FL).
no code implementations • Findings (ACL) 2022 • Khalil Mrini, Shaoliang Nie, Jiatao Gu, Sinong Wang, Maziar Sanjabi, Hamed Firooz
Without the use of a knowledge base or candidate sets, our model sets a new state of the art in two benchmark datasets of entity linking: COMETA in the biomedical domain, and AIDA-CoNLL in the news domain.
2 code implementations • 8 Apr 2022 • Krishna Pillutla, Kshitiz Malik, Abdelrahman Mohamed, Michael Rabbat, Maziar Sanjabi, Lin Xiao
We consider two federated learning algorithms for training partially personalized models, where the shared and personal parameters are updated either simultaneously or alternately on the devices.
no code implementations • 3 Mar 2022 • Neha Kalibhat, Kanika Narang, Hamed Firooz, Maziar Sanjabi, Soheil Feizi
Fine-tuning with Q-Score regularization can boost the linear probing accuracy of SSL models by up to 5. 8% on ImageNet-100 and 3. 7% on ImageNet-1K compared to their baselines.
no code implementations • 31 Dec 2021 • Nimit S. Sohoni, Maziar Sanjabi, Nicolas Ballas, Aditya Grover, Shaoliang Nie, Hamed Firooz, Christopher Ré
Theoretically, we provide generalization bounds for our approach in terms of the worst-group performance, which scale with respect to both the total number of training points and the number of training points with group labels.
no code implementations • 27 Dec 2021 • Ajinkya Tejankar, Maziar Sanjabi, Bichen Wu, Saining Xie, Madian Khabsa, Hamed Pirsiavash, Hamed Firooz
In this paper, we focus on teasing out what parts of the language supervision are essential for training zero-shot image classification models.
1 code implementation • BigScience (ACL) 2022 • Aaron Chan, Maziar Sanjabi, Lambert Mathias, Liang Tan, Shaoliang Nie, Xiaochang Peng, Xiang Ren, Hamed Firooz
An extractive rationale explains a language model's (LM's) prediction on a given task instance by highlighting the text inputs that most influenced the prediction.
1 code implementation • 13 Sep 2021 • Tian Li, Ahmad Beirami, Maziar Sanjabi, Virginia Smith
Finally, we demonstrate that TERM can be used for a multitude of applications in machine learning, such as enforcing fairness between subgroups, mitigating the effect of outliers, and handling class imbalance.
no code implementations • Findings (EMNLP) 2021 • Woojeong Jin, Maziar Sanjabi, Shaoliang Nie, Liang Tan, Xiang Ren, Hamed Firooz
The idea aims at mimicking a teacher's modality-specific predictions by introducing auxiliary loss terms for each modality.
no code implementations • 8 Sep 2020 • Tianjian Huang, Prajwal Singhania, Maziar Sanjabi, Pabitra Mitra, Meisam Razaviyayn
For such optimization problems, we study the performance of the Alternating Direction Method of Multipliers for Quantization ($\texttt{ADMM-Q}$) algorithm, which is a variant of the widely-used ADMM method applied to our discrete optimization problem.
2 code implementations • ICLR 2021 • Tian Li, Ahmad Beirami, Maziar Sanjabi, Virginia Smith
Empirical risk minimization (ERM) is typically designed to perform well on the average loss, which can result in estimators that are sensitive to outliers, generalize poorly, or treat subgroups unfairly.
no code implementations • 15 Jun 2020 • Meisam Razaviyayn, Tianjian Huang, Songtao Lu, Maher Nouiehed, Maziar Sanjabi, Mingyi Hong
The min-max optimization problem, also known as the saddle point problem, is a classical optimization problem which is also studied in the context of zero-sum games.
2 code implementations • 7 Jan 2020 • Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, Virginia Smith
Federated learning aims to jointly learn statistical models over massively distributed remote devices.
no code implementations • 22 Nov 2019 • Maziar Sanjabi, Sina Baharlouei, Meisam Razaviyayn, Jason D. Lee
We study the optimization problem for decomposing $d$ dimensional fourth-order Tensors with $k$ non-orthogonal components.
2 code implementations • ICLR 2020 • Tian Li, Maziar Sanjabi, Ahmad Beirami, Virginia Smith
Federated learning involves training statistical models in massive, heterogeneous networks.
2 code implementations • 22 Apr 2019 • Babak Barazandeh, Meisam Razaviyayn, Maziar Sanjabi
This design helps us to avoid the min-max formulation and leads to an optimization problem that is stable and could be solved efficiently.
1 code implementation • NeurIPS 2019 • Maher Nouiehed, Maziar Sanjabi, Tianjian Huang, Jason D. Lee, Meisam Razaviyayn
In this paper, we study the problem in the non-convex regime and show that an \varepsilon--first order stationary point of the game can be computed when one of the player's objective can be optimized to global optimality efficiently.
19 code implementations • 14 Dec 2018 • Tian Li, Anit Kumar Sahu, Manzil Zaheer, Maziar Sanjabi, Ameet Talwalkar, Virginia Smith
Theoretically, we provide convergence guarantees for our framework when learning over data from non-identical distributions (statistical heterogeneity), and while adhering to device-level systems constraints by allowing each participating device to perform a variable amount of work (systems heterogeneity).
no code implementations • 7 Dec 2018 • Maziar Sanjabi, Meisam Razaviyayn, Jason D. Lee
In this short note, we consider the problem of solving a min-max zero-sum game.
no code implementations • NeurIPS 2018 • Maziar Sanjabi, Jimmy Ba, Meisam Razaviyayn, Jason D. Lee
A popular GAN formulation is based on the use of Wasserstein distance as a metric between probability distributions.
2 code implementations • NeurIPS 2017 • Virginia Smith, Chao-Kai Chiang, Maziar Sanjabi, Ameet Talwalkar
Federated learning poses new statistical and systems challenges in training machine learning models over distributed networks of devices.