Publications

Systematic Rectification of Language Models via Dead-end Analysis
Meng Cao
Mehdi Fatemi
Jackie Cheung
Samira Shabanian
With adversarial or otherwise normal prompts, existing large language models (LLM) can be pushed to generate toxic discourses. One way to re… (see more)duce the risk of LLMs generating undesired discourses is to alter the training of the LLM. This can be very restrictive due to demanding computation requirements. Other methods rely on rule-based or prompt-based token elimination, which are limited as they dismiss future tokens and the overall meaning of the complete discourse. Here, we center detoxification on the probability that the finished discourse is ultimately considered toxic. That is, at each point, we advise against token selections proportional to how likely a finished text from this point will be toxic. To this end, we formally extend the dead-end theory from the recent reinforcement learning (RL) literature to also cover uncertain outcomes. Our approach, called rectification, utilizes a separate but significantly smaller model for detoxification, which can be applied to diverse LLMs as long as they share the same vocabulary. Importantly, our method does not require access to the internal representations of the LLM, but only the token probability distribution at each decoding step. This is crucial as many LLMs today are hosted in servers and only accessible through APIs. When applied to various LLMs, including GPT-3, our approach significantly improves the generated discourse compared to the base LLMs and other techniques in terms of both the overall language and detoxification performance.
2023-02-01
ICLR.cc/2023/Conference (poster)
doi.org
openreview.net
The clinical value of Aspergillus-specific IgG antibody test in the diagnosis of nonneutropenic invasive pulmonary aspergillosis.
Yajie Lu
Lulu Liu
Hongxing Li
Bilin Chen
Yu-hui Gu
Li Wang
Chunlai Feng
Cheng Chen
Yanbin Chen
Wenkui Sun
X. Cui
Min Cao
Yujian Tao
Jinjin Zhong
Huanhuan Zhong
Yueyan Ni
Yuchen Cai
M. Song
X. Liu
Yi Shi Li Liu … (see 1 more)
Xin Su
2023-02-01
Clinical Microbiology and Infection (published)
doi.org
The Hidden Uniform Cluster Prior in Self-Supervised Learning
Mahmoud Assran
Randall Balestriero
Quentin Duval
Florian Bordes
Ishan Misra
Piotr Bojanowski
Pascal Vincent
Michael Rabbat
Nicolas Ballas
A successful paradigm in representation learning is to perform self-supervised pretraining using tasks based on mini-batch statistics (e.g.,… (see more) SimCLR, VICReg, SwAV, MSN). We show that in the formulation of all these methods is an overlooked prior to learn features that enable uniform clustering of the data. While this prior has led to remarkably semantic representations when pretraining on class-balanced data, such as ImageNet, we demonstrate that it can hamper performance when pretraining on class-imbalanced data. By moving away from conventional uniformity priors and instead preferring power-law distributed feature clusters, we show that one can improve the quality of the learned representations on real-world class-imbalanced datasets. To demonstrate this, we develop an extension of the Masked Siamese Networks (MSN) method to support the use of arbitrary features priors.
2023-02-01
ICLR.cc/2023/Conference (poster)
doi.org
openreview.net
Understanding Zero-shot Adversarial Robustness for Large-Scale Models
Chengzhi Mao
Scott Geng
Junfeng Yang
Xin Wang
Carl Vondrick
Pretrained large-scale vision-language models like CLIP have exhibited strong generalization over unseen tasks. Yet imperceptible adversaria… (see more)l perturbations can significantly reduce CLIP's performance on new tasks. In this work, we identify and explore the problem of adapting large-scale models for zero-shot adversarial robustness. We first identify two key factors during model adaption--training losses and adaptation methods--that affect the model's zero-shot adversarial robustness. We then propose a text-guided contrastive adversarial training loss, which aligns the text embeddings and the adversarial visual features with contrastive learning on a small set of training data. We apply this training loss to two adaption methods, model finetuning and visual prompt tuning. We find that visual prompt tuning is more effective in the absence of texts, while finetuning wins in the existence of text guidance. Overall, our approach significantly improves the zero-shot adversarial robustness over CLIP, seeing an average improvement of 31 points over ImageNet and 15 zero-shot datasets. We hope this work can shed light on understanding the zero-shot adversarial robustness of large-scale models.
2023-02-01
ICLR.cc/2023/Conference (poster)
doi.org
openreview.net
A Unified Approach to Reinforcement Learning, Quantal Response Equilibria, and Two-Player Zero-Sum Games
Samuel Sokota
Ryan D'Orazio
J Zico Kolter
Nicolas Loizou
Marc Lanctot
Ioannis Mitliagkas
Noam Brown
Christian Kroer
2023-02-01
ICLR.cc/2023/Conference (poster)
openreview.net
openreview.net
Unmasking the Lottery Ticket Hypothesis: What's Encoded in a Winning Ticket's Mask?
Mansheej Paul
Feng Chen
Brett W. Larsen
Jonathan Frankle
Surya Ganguli
Gintare Karolina Dziugaite
Modern deep learning involves training costly, highly overparameterized networks, thus motivating the search for sparser networks that can s… (see more)till be trained to the same accuracy as the full network (i.e. matching). Iterative magnitude pruning (IMP) is a state of the art algorithm that can find such highly sparse matching subnetworks, known as winning tickets. IMP operates by iterative cycles of training, masking smallest magnitude weights, rewinding back to an early training point, and repeating. Despite its simplicity, the underlying principles for when and how IMP finds winning tickets remain elusive. In particular, what useful information does an IMP mask found at the end of training convey to a rewound network near the beginning of training? How does SGD allow the network to extract this information? And why is iterative pruning needed? We develop answers in terms of the geometry of the error landscape. First, we find that
2023-02-01
ICLR.cc/2023/Conference (notable)
doi.org
openreview.net
Where to Begin? On the Impact of Pre-Training and Initialization in Federated Learning
John Nguyen
Jianyu Wang
Kshitiz Malik
Maziar Sanjabi
Michael Rabbat
AI Meta
2023-02-01
ICLR.cc/2023/Conference (notable)
doi.org
openreview.net
BARVINN: Arbitrary Precision DNN Accelerator Controlled by a RISC-V CPU
Mohammadhossein Askarihemmat
Sean Wagner
Olexa Bilaniuk
Yassine Hariri
Yvon Savaria
Jean-Pierre David
2023-01-31
Proceedings of the 28th Asia and South Pacific Design Automation Conference (published)
doi.org
arxiv.org
Graph-based Time-Series Anomaly Detection: A Survey
Thi Kieu Khanh Ho
Ali Karami
Narges Armanfard
With the recent advances in technology, a wide range of systems continue to collect a large amount of data over time and thus generate time … (see more)series. Time-Series Anomaly Detection (TSAD) is an important task in various time-series applications such as e-commerce, cybersecurity, vehicle maintenance, and healthcare monitoring. However, this task is very challenging as it requires considering both the intra-variable dependency and the inter-variable dependency, where a variable can be defined as an observation in time series data. Recent graph-based approaches have made impressive progress in tackling the challenges of this field. In this survey, we conduct a comprehensive and up-to-date review of Graph-based TSAD (G-TSAD). First, we explore the significant potential of graph representation learning for time-series data. Then, we review state-of-the-art graph anomaly detection techniques in the context of time series and discuss their strengths and drawbacks. Finally, we discuss the technical challenges and potential future directions for possible improvements in this research field.
2023-01-31
ArXiv (preprint)
doi.org
arxiv.org
Single-cell multi-omic topic embedding reveals cell-type-specific and COVID-19 severity-related immune signatures
Manqi Zhou
Hao Zhang
Zilong Bai
Dylan Mann-Krzisnik
Fei Wang
Yue Li
The advent of single-cell multi-omics sequencing technology makes it possible for re-searchers to leverage multiple modalities for individua… (see more)l cells and explore cell heterogeneity. However, the high dimensional, discrete, and sparse nature of the data make the downstream analysis particularly challenging. Most of the existing computational methods for single-cell data analysis are either limited to single modality or lack flexibility and interpretability. In this study, we propose an interpretable deep learning method called multi-omic embedded topic model (moETM) to effectively perform integrative analysis of high-dimensional single-cell multimodal data. moETM integrates multiple omics data via a product-of-experts in the encoder for efficient variational inference and then employs multiple linear decoders to learn the multi-omic signatures of the gene regulatory programs. Through comprehensive experiments on public single-cell transcriptome and chromatin accessibility data (i.e., scRNA+scATAC), as well as scRNA and proteomic data (i.e., CITE-seq), moETM demonstrates superior performance compared with six state-of-the-art single-cell data analysis methods on seven publicly available datasets. By applying moETM to the scRNA+scATAC data in human bone marrow mononuclear cells (BMMCs), we identified sequence motifs corresponding to the transcription factors that regulate immune gene signatures. Applying moETM analysis to CITE-seq data from the COVID-19 patients revealed not only known immune cell-type-specific signatures but also composite multi-omic biomarkers of critical conditions due to COVID-19, thus providing insights from both biological and clinical perspectives.
2023-01-31
bioRxiv (preprint)
doi.org
Technical Note—Risk-Averse Regret Minimization in Multistage Stochastic Programs
Mehran Poursoltani
Érick Delage
Angelos Georghiou
2023-01-30
Operational Research (published)
doi.org
Leveraging the Third Dimension in Contrastive Learning
Sumukh K Aithal
Anirudh Goyal
Alex Lamb
Yoshua Bengio
Michael Curtis Mozer
Self-Supervised Learning (SSL) methods operate on unlabeled data to learn robust representations useful for downstream tasks. Most SSL metho… (see more)ds rely on augmentations obtained by transforming the 2D image pixel map. These augmentations ignore the fact that biological vision takes place in an immersive three-dimensional, temporally contiguous environment, and that low-level biological vision relies heavily on depth cues. Using a signal provided by a pretrained state-of-the-art monocular RGB-to-depth model (the \emph{Depth Prediction Transformer}, Ranftl et al., 2021), we explore two distinct approaches to incorporating depth signals into the SSL framework. First, we evaluate contrastive learning using an RGB+depth input representation. Second, we use the depth signal to generate novel views from slightly different camera positions, thereby producing a 3D augmentation for contrastive learning. We evaluate these two approaches on three different SSL methods -- BYOL, SimSiam, and SwAV -- using ImageNette (10 class subset of ImageNet), ImageNet-100 and ImageNet-1k datasets. We find that both approaches to incorporating depth signals improve the robustness and generalization of the baseline SSL methods, though the first approach (with depth-channel concatenation) is superior. For instance, BYOL with the additional depth channel leads to an increase in downstream classification accuracy from 85.3\% to 88.0\% on ImageNette and 84.1\% to 87.0\% on ImageNet-C.
2023-01-27
ArXiv (preprint)
doi.org
openreview.net