Self-supervised instance discrimination is an effective contrastive pretext task to learn feature representations and address limited medical image annotations. The idea is to make features of transformed versions of the same images similar while forcing all other augmented images' representations to contrast. However, this instance-based contrastive learning leaves performance on the table by failing to maximize feature affinity between images with similar content while counter-productively pushing their representations apart. Recent improvements on this paradigm (e.g., leveraging multi-modal data, different images in longitudinal studies, spatial correspondences) either relied on additional views or made stringent assumptions about data properties, which can sacrifice generalizability and applicability. To address this challenge, we propose a new self-supervised contrastive learning method that uses unsupervised feature clustering to better select positive and negative image samples. More specifically, we produce pseudo-classes by hierarchically clustering features obtained by an auto-encoder in an unsupervised manner, and prevent destructive interference during contrastive learning by avoiding the selection of negatives from the same pseudo-class. Experiments on 2D skin dermoscopic image segmentation and 3D multi-class whole heart CT segmentation demonstrate that our method outperforms state-of-the-art self-supervised contrastive techniques on these tasks.
Deep anomaly detection (AD) aims to provide robust and efficient classifiers for one-class and unbalanced settings. However current AD models still struggle on edge-case normal samples and are often unable to keep high performance over different scales of anomalies. Moreover, there currently does not exist a unified framework efficiently covering both one-class and unbalanced learnings. In the light of these limitations, we introduce a new two-stage anomaly detector which memorizes during training multi-scale normal prototypes to compute an anomaly deviation score. First, we simultaneously learn representations and memory modules on multiple scales using a novel memory-augmented contrastive learning. Then, we train an anomaly distance detector on the spatial deviation maps between prototypes and observations. Our model highly improves the state-of-the-art performance on a wide range of object, style and local anomalies with up to 50% error relative improvement on CIFAR-100. It is also the first model to keep high performance across the one-class and unbalanced settings.
Observing a human demonstrator manipulate objects provides a rich, scalable and inexpensive source of data for learning robotic policies. However, transferring skills from human videos to a robotic manipulator poses several challenges, not least a difference in action and observation spaces. In this work, we use unlabeled videos of humans solving a wide range of manipulation tasks to learn a task-agnostic reward function for robotic manipulation policies. Thanks to the diversity of this training data, the learned reward function sufficiently generalizes to image observations from a previously unseen robot embodiment and environment to provide a meaningful prior for directed exploration in reinforcement learning. We propose two methods for scoring states relative to a goal image: through direct temporal regression, and through distances in an embedding space obtained with time-contrastive learning. By conditioning the function on a goal image, we are able to reuse one model across a variety of tasks. Unlike prior work on leveraging human videos to teach robots, our method, Human Offline Learned Distances (HOLD) requires neither a priori data from the robot environment, nor a set of task-specific human demonstrations, nor a predefined notion of correspondence across morphologies, yet it is able to accelerate training of several manipulation tasks on a simulated robot arm compared to using only a sparse reward obtained from task completion.
This technical report describes the CONE approach for Ego4D Natural Language Queries (NLQ) Challenge in ECCV 2022. We leverage our model CONE, an efficient window-centric COarse-to-fiNE alignment framework. Specifically, CONE dynamically slices the long video into candidate windows via a sliding window approach. Centering at windows, CONE (1) learns the inter-window (coarse-grained) semantic variance through contrastive learning and speeds up inference by pre-filtering the candidate windows relevant to the NL query, and (2) conducts intra-window (fine-grained) candidate moments ranking utilizing the powerful multi-modal alignment ability of the contrastive vision-text pre-trained model EgoVLP. On the blind test set, CONE achieves 15.26 and 9.24 for R1@IoU=0.3 and R1@IoU=0.5, respectively.
Satellite imagery is being leveraged for many societally critical tasks across climate, economics, and public health. Yet, because of heterogeneity in landscapes (e.g. how a road looks in different places), models can show disparate performance across geographic areas. Given the important potential of disparities in algorithmic systems used in societal contexts, here we consider the risk of urban-rural disparities in identification of land-cover features. This is via semantic segmentation (a common computer vision task in which image regions are labelled according to what is being shown) which uses pre-trained image representations generated via contrastive self-supervised learning. We propose fair dense representation with contrastive learning (FairDCL) as a method for de-biasing the multi-level latent space of convolution neural network models. The method improves feature identification by removing spurious model representations which are disparately distributed across urban and rural areas, and is achieved in an unsupervised way by contrastive pre-training. The obtained image representation mitigates downstream urban-rural prediction disparities and outperforms state-of-the-art baselines on real-world satellite images. Embedding space evaluation and ablation studies further demonstrate FairDCL's robustness. As generalizability and robustness in geographic imagery is a nascent topic, our work motivates researchers to consider metrics beyond average accuracy in such applications.
Understanding of spatial attributes is central to effective 3D radiology image analysis where crop-based learning is the de facto standard. Given an image patch, its core spatial properties (e.g., position & orientation) provide helpful priors on expected object sizes, appearances, and structures through inherent anatomical consistencies. Spatial correspondences, in particular, can effectively gauge semantic similarities between inter-image regions, while their approximate extraction requires no annotations or overbearing computational costs. However, recent 3D contrastive learning approaches either neglect correspondences or fail to maximally capitalize on them. To this end, we propose an extensible 3D contrastive framework (Spade, for Spatial Debiasing) that leverages extracted correspondences to select more effective positive & negative samples for representation learning. Our method learns both globally invariant and locally equivariant representations with downstream segmentation in mind. We also propose separate selection strategies for global & local scopes that tailor to their respective representational requirements. Compared to recent state-of-the-art approaches, Spade shows notable improvements on three downstream segmentation tasks (CT Abdominal Organ, CT Heart, MR Heart).
In this paper, to unveil interpretable development-specific gene signatures in human PFC, we propose a novel gene selection method, named Interpretable Causality Gene Selection (ICGS), which adopts a Bayesian Network (BN) to represent causality between multiple gene variables and a development variable. The proposed ICGS method combines the positive instances-based contrastive learning with a Variational AutoEncoder (VAE) to obtain this optimal BN structure and use a Markov Blanket (MB) to identify gene signatures causally related to the development variable. Moreover, the differential expression genes (DEGs) are used to filter redundant genes before gene selection. In order to identify gene signatures, we apply the proposed ICGS to the human PFC single-cell transcriptomics data. The experimental results demonstrate that the proposed method can effectively identify interpretable development-specific gene signatures in human PFC. Gene ontology enrichment analysis and ASD-related gene analysis show that these identified gene signatures reveal the key biological processes and pathways in human PFC and have more potential for neurodevelopment disorder cure. These gene signatures are expected to bring important implications for understanding PFC development heterogeneity and function in humans.
Generative models have demonstrated impressive results on Aspect-based Sentiment Analysis (ABSA) tasks, particularly for the emerging task of extracting Aspect-Category-Opinion-Sentiment (ACOS) quadruples. However, these models struggle with implicit sentiment expressions, which are commonly observed in opinionated content such as online reviews. In this work, we introduce GEN-SCL-NAT, which consists of two techniques for improved structured generation for ACOS quadruple extraction. First, we propose GEN-SCL, a supervised contrastive learning objective that aids quadruple prediction by encouraging the model to produce input representations that are discriminable across key input attributes, such as sentiment polarity and the existence of implicit opinions and aspects. Second, we introduce GEN-NAT, a new structured generation format that better adapts autoregressive encoder-decoder models to extract quadruples in a generative fashion. Experimental results show that GEN-SCL-NAT achieves top performance across three ACOS datasets, averaging 1.48% F1 improvement, with a maximum 1.73% increase on the LAPTOP-L1 dataset. Additionally, we see significant gains on implicit aspect and opinion splits that have been shown as challenging for existing ACOS approaches.
Building accurate Deep Learning (DL) models for brain age prediction is a very relevant topic in neuroimaging, as it could help better understand neurodegenerative disorders and find new biomarkers. To estimate accurate and generalizable models, large datasets have been collected, which are often multi-site and multi-scanner. This large heterogeneity negatively affects the generalization performance of DL models since they are prone to overfit site-related noise. Recently, contrastive learning approaches have been shown to be more robust against noise in data or labels. For this reason, we propose a novel contrastive learning regression loss for robust brain age prediction using MRI scans. Our method achieves state-of-the-art performance on the OpenBHB challenge, yielding the best generalization capability and robustness to site-related noise.
Leveraging the fact that speaker identity and content vary on different time scales, \acrlong{fhvae} (\acrshort{fhvae}) uses different latent variables to symbolize these two attributes. Disentanglement of these attributes is carried out by different prior settings of the corresponding latent variables. For the prior of speaker identity variable, \acrshort{fhvae} assumes it is a Gaussian distribution with an utterance-scale varying mean and a fixed variance. By setting a small fixed variance, the training process promotes identity variables within one utterance gathering close to the mean of their prior. However, this constraint is relatively weak, as the mean of the prior changes between utterances. Therefore, we introduce contrastive learning into the \acrshort{fhvae} framework, to make the speaker identity variables gathering when representing the same speaker, while distancing themselves as far as possible from those of other speakers. The model structure has not been changed in this work but only the training process, thus no additional cost is needed during testing. Voice conversion has been chosen as the application in this paper. Latent variable evaluations include speaker verification and identification for the speaker identity variable, and speech recognition for the content variable. Furthermore, assessments of voice conversion performance are on the grounds of fake speech detection experiments. Results show that the proposed method improves both speaker identity and content feature extraction compared to \acrshort{fhvae}, and has better performance than baseline on conversion.
Finding informative low-dimensional representations that can be computed efficiently in large datasets is an important problem in data analysis. Recently, contrastive Principal Component Analysis (cPCA) was proposed as a more informative generalization of PCA that takes advantage of contrastive learning. However, the performance of cPCA is sensitive to hyper-parameter choice and there is currently no online algorithm for implementing cPCA. Here, we introduce a modified cPCA method, which we denote cPCA*, that is more interpretable and less sensitive to the choice of hyper-parameter. We derive an online algorithm for cPCA* and show that it maps onto a neural network with local learning rules, so it can potentially be implemented in energy efficient neuromorphic hardware. We evaluate the performance of our online algorithm on real datasets and highlight the differences and similarities with the original formulation.
Given the fact description text of a legal case, legal judgment prediction (LJP) aims to predict the case's charge, law article and penalty term. A core problem of LJP is how to distinguish confusing legal cases, where only subtle text differences exist. Previous studies fail to distinguish different classification errors with a standard cross-entropy classification loss, and ignore the numbers in the fact description for predicting the term of penalty. To tackle these issues, in this work, first, we propose a moco-based supervised contrastive learning to learn distinguishable representations, and explore the best strategy to construct positive example pairs to benefit all three subtasks of LJP simultaneously. Second, in order to exploit the numbers in legal cases for predicting the penalty terms of certain cases, we further enhance the representation of the fact description with extracted crime amounts which are encoded by a pre-trained numeracy model. Extensive experiments on public benchmarks show that the proposed method achieves new state-of-the-art results, especially on confusing legal cases. Ablation studies also demonstrate the effectiveness of each component.
Contrastive learning (CL) pre-trains general-purpose encoders using an unlabeled pre-training dataset, which consists of images or image-text pairs. CL is vulnerable to data poisoning based backdoor attacks (DPBAs), in which an attacker injects poisoned inputs into the pre-training dataset so the encoder is backdoored. However, existing DPBAs achieve limited effectiveness. In this work, we take the first step to analyze the limitations of existing backdoor attacks and propose new DPBAs called CorruptEncoder to CL. CorruptEncoder introduces a new attack strategy to create poisoned inputs and uses a theory-guided method to maximize attack effectiveness. Our experiments show that CorruptEncoder substantially outperforms existing DPBAs. In particular, CorruptEncoder is the first DPBA that achieves more than 90% attack success rates with only a few (3) reference images and a small poisoning ratio 0.5%. Moreover, we also propose a defense, called localized cropping, to defend against DPBAs. Our results show that our defense can reduce the effectiveness of DPBAs, but it sacrifices the utility of the encoder, highlighting the need for new defenses.
The retinogeniculate pathway (RGVP) is responsible for carrying visual information from the retina to the lateral geniculate nucleus. Identification and visualization of the RGVP are important in studying the anatomy of the visual system and can inform treatment of related brain diseases. Diffusion MRI (dMRI) tractography is an advanced imaging method that uniquely enables in vivo mapping of the 3D trajectory of the RGVP. Currently, identification of the RGVP from tractography data relies on expert (manual) selection of tractography streamlines, which is time-consuming, has high clinical and expert labor costs, and affected by inter-observer variability. In this paper, we present what we believe is the first deep learning framework, namely DeepRGVP, to enable fast and accurate identification of the RGVP from dMRI tractography data. We design a novel microstructure-informed supervised contrastive learning method that leverages both streamline label and tissue microstructure information to determine positive and negative pairs. We propose a simple and successful streamline-level data augmentation method to address highly imbalanced training data, where the number of RGVP streamlines is much lower than that of non-RGVP streamlines. We perform comparisons with several state-of-the-art deep learning methods that were designed for tractography parcellation, and we show superior RGVP identification results using DeepRGVP.
The existing SSCL of RSI is built based on constructing positive and negative sample pairs. However, due to the richness of RSI ground objects and the complexity of the RSI contextual semantics, the same RSI patches have the coexistence and imbalance of positive and negative samples, which causing the SSCL pushing negative samples far away while pushing positive samples far away, and vice versa. We call this the sample confounding issue (SCI). To solve this problem, we propose a False negAtive sampLes aware contraStive lEarning model (FALSE) for the semantic segmentation of high-resolution RSIs. Since the SSCL pretraining is unsupervised, the lack of definable criteria for false negative sample (FNS) leads to theoretical undecidability, we designed two steps to implement the FNS approximation determination: coarse determination of FNS and precise calibration of FNS. We achieve coarse determination of FNS by the FNS self-determination (FNSD) strategy and achieve calibration of FNS by the FNS confidence calibration (FNCC) loss function. Experimental results on three RSI semantic segmentation datasets demonstrated that the FALSE effectively improves the accuracy of the downstream RSI semantic segmentation task compared with the current three models, which represent three different types of SSCL models. The mean Intersection-over-Union on ISPRS Potsdam dataset is improved by 0.7\% on average; on CVPR DGLC dataset is improved by 12.28\% on average; and on Xiangtan dataset this is improved by 1.17\% on average. This indicates that the SSCL model has the ability to self-differentiate FNS and that the FALSE effectively mitigates the SCI in self-supervised contrastive learning. The source code is available at https://github.com/GeoX-Lab/FALSE.
Inspired by the curvature of space-time (Einstein, 1921), we introduce Curved Contrastive Learning (CCL), a novel representation learning technique for learning the relative turn distance between utterance pairs in multi-turn dialogues. The resulting bi-encoder models can guide transformers as a response ranking model towards a goal in a zero-shot fashion by projecting the goal utterance and the corresponding reply candidates into a latent space. Here the cosine similarity indicates the distance/reachability of a candidate utterance toward the corresponding goal. Furthermore, we explore how these forward-entailing language representations can be utilized for assessing the likelihood of sequences by the entailment strength i.e. through the cosine similarity of its individual members (encoded separately) as an emergent property in the curved space. These non-local properties allow us to imagine the likelihood of future patterns in dialogues, specifically by ordering/identifying future goal utterances that are multiple turns away, given a dialogue context. As part of our analysis, we investigate characteristics that make conversations (un)plannable and find strong evidence of planning capability over multiple turns (in 61.56% over 3 turns) in conversations from the DailyDialog (Li et al., 2017) dataset. Finally, we show how we achieve higher efficiency in sequence modeling tasks compared to previous work thanks to our relativistic approach, where only the last utterance needs to be encoded and computed during inference.
In this work, we present a conceptually simple and effective method to train a strong bilingual/multilingual multimodal representation model. Starting from the pre-trained multimodal representation model CLIP released by OpenAI, we altered its text encoder with a pre-trained multilingual text encoder XLM-R, and aligned both languages and image representations by a two-stage training schema consisting of teacher learning and contrastive learning. We validate our method through evaluations of a wide range of tasks. We set new state-of-the-art performances on a bunch of tasks including ImageNet-CN, Flicker30k-CN, COCO-CN and XTD. Further, we obtain very close performances with CLIP on almost all tasks, suggesting that one can simply alter the text encoder in CLIP for extended capabilities such as multilingual understanding. Our models and code are available at https://github.com/FlagAI-Open/FlagAI.
The prevalence of abusive language on different online platforms has been a major concern that raises the need for automated cross-platform abusive language detection. However, prior works focus on concatenating data from multiple platforms, inherently adopting Empirical Risk Minimization (ERM) method. In this work, we address this challenge from the perspective of domain generalization objective. We design SCL-Fish, a supervised contrastive learning integrated meta-learning algorithm to detect abusive language on unseen platforms. Our experimental analysis shows that SCL-Fish achieves better performance over ERM and the existing state-of-the-art models. We also show that SCL-Fish is data-efficient and achieves comparable performance with the large-scale pre-trained models upon finetuning for the abusive language detection task.
Climate models often require post-processing in order to make accurate estimates of local climate risk. The most common post-processing applied is bias-correction and spatial resolution enhancement. However, the statistical methods typically used for this not only are incapable of capturing multivariate spatial correlation information but are also reliant on rich observational data often not available outside of developed countries, limiting their potential. Here we propose an alternative approach to this challenge based on a combination of image super resolution (SR) and contrastive learning generative adversarial networks (GANs). We benchmark performance against NASA's flagship post-processed CMIP6 climate model product, NEX-GDDP. We find that our model successfully reaches a spatial resolution double that of NASA's product while also achieving comparable or improved levels of bias correction in both daily precipitation and temperature. The resulting higher fidelity simulations of present and forward-looking climate can enable more local, accurate models of hazards like flooding, drought, and heatwaves.
Clustering is the task of gathering similar data samples into clusters without using any predefined labels. It has been widely studied in machine learning literature, and recent advancements in deep learning have revived interest in this field. Contrastive clustering (CC) models are a staple of deep clustering in which positive and negative pairs of each data instance are generated through data augmentation. CC models aim to learn a feature space where instance-level and cluster-level representations of positive pairs are grouped together. Despite improving the SOTA, these algorithms ignore the cross-instance patterns, which carry essential information for improving clustering performance. This increases the false-negative-pair rate of the model while decreasing its true-positive-pair rate. In this paper, we propose a novel contrastive clustering method, Cross-instance guided Contrastive Clustering (C3), that considers the cross-sample relationships to increase the number of positive pairs and mitigate the impact of false negative, noise, and anomaly sample on the learned representation of data. In particular, we define a new loss function that identifies similar instances using the instance-level representation and encourages them to aggregate together. Moreover, we propose a novel weighting method to select negative samples in a more efficient way. Extensive experimental evaluations show that our proposed method can outperform state-of-the-art algorithms on benchmark computer vision datasets: we improve the clustering accuracy by 6.6%, 3.3%, 5.0%, 1.3% and 0.3% on CIFAR-10, CIFAR-100, ImageNet-10, ImageNet-Dogs, and Tiny-ImageNet.