Abstract
Molecular representation learning is crucial for the problem of molecular property prediction, where graph neural networks (GNNs) serve as an effective solution due to their structure modeling capabilities. Since labeled data is often scarce and expensive to obtain, it is a great challenge for GNNs to generalize in the extensive molecular space. Recently, the training paradigm of "pre-train, fine-tune" has been leveraged to improve the generalization capabilities of GNNs. It uses self-supervised information to pre-train the GNN, and then performs fine-tuning to optimize the downstream task with just a few labels. However, pre-training does not always yield statistically significant improvement, especially for self-supervised learning with random structural masking. In fact, the molecular structure is characterized by motif subgraphs, which are frequently occurring and influence molecular properties. To leverage the task-related motifs, we propose a novel paradigm of "pre-train, prompt, fine-tune" for molecular representation learning, named molecule continuous prompt tuning (MolCPT). MolCPT defines a motif prompting function that uses the pre-trained model to project the standalone input into an expressive prompt. The prompt effectively augments the molecular graph with meaningful motifs in the continuous representation space; this provides more structural patterns to aid the downstream classifier in identifying molecular properties. Extensive experiments on several benchmark datasets show that MolCPT efficiently generalizes pre-trained GNNs for molecular property prediction, with or without a few fine-tuning steps.
Abstract
Relation extraction (RE), which has relied on structurally annotated corpora for model training, has been particularly challenging in low-resource scenarios and domains. Recent literature has tackled low-resource RE by self-supervised learning, where the solution involves pretraining the relation embedding by RE-based objective and finetuning on labeled data by classification-based objective. However, a critical challenge to this approach is the gap in objectives, which prevents the RE model from fully utilizing the knowledge in pretrained representations. In this paper, we aim at bridging the gap and propose to pretrain and finetune the RE model using consistent objectives of contrastive learning. Since in this kind of representation learning paradigm, one relation may easily form multiple clusters in the representation space, we further propose a multi-center contrastive loss that allows one relation to form multiple clusters to better align with pretraining. Experiments on two document-level RE datasets, BioRED and Re-DocRED, demonstrate the effectiveness of our method. Particularly, when using 1% end-task training data, our method outperforms PLM-based RE classifier by 10.5% and 5.8% on the two datasets, respectively.
Joint Embedding of 2D and 3D Networks for Medical Image Anomaly Detection
Authors: Inha Kang, Jinah Park
Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Obtaining ground truth data in medical imaging has difficulties due to the fact that it requires a lot of annotating time from the experts in the field. Also, when trained with supervised learning, it detects only the cases included in the labels. In real practice, we want to also open to other possibilities than the named cases while examining the medical images. As a solution, the need for anomaly detection that can detect and localize abnormalities by learning the normal characteristics using only normal images is emerging. With medical image data, we can design either 2D or 3D networks of self-supervised learning for anomaly detection task. Although 3D networks, which learns 3D structures of the human body, show good performance in 3D medical image anomaly detection, they cannot be stacked in deeper layers due to memory problems. While 2D networks have advantage in feature detection, they lack 3D context information. In this paper, we develop a method for combining the strength of the 3D network and the strength of the 2D network through joint embedding. We also propose the pretask of self-supervised learning to make it possible for the networks to learn efficiently. Through the experiments, we show that the proposed method achieves better performance in both classification and segmentation tasks compared to the SoTA method.
TruFor: Leveraging all-round clues for trustworthy image forgery detection and localization
Abstract
In this paper we present TruFor, a forensic framework that can be applied to a large variety of image manipulation methods, from classic cheapfakes to more recent manipulations based on deep learning. We rely on the extraction of both high-level and low-level traces through a transformer-based fusion architecture that combines the RGB image and a learned noise-sensitive fingerprint. The latter learns to embed the artifacts related to the camera internal and external processing by training only on real data in a self-supervised manner. Forgeries are detected as deviations from the expected regular pattern that characterizes each pristine image. Looking for anomalies makes the approach able to robustly detect a variety of local manipulations, ensuring generalization. In addition to a pixel-level localization map and a whole-image integrity score, our approach outputs a reliability map that highlights areas where localization predictions may be error-prone. This is particularly important in forensic applications in order to reduce false alarms and allow for a large scale analysis. Extensive experiments on several datasets show that our method is able to reliably detect and localize both cheapfakes and deepfakes manipulations outperforming state-of-the-art works. Code will be publicly available at https://grip-unina.github.io/TruFor/
THMA: Tencent HD Map AI System for Creating HD Map Annotations
Authors: Kun Tang, Xu Cao, Zhipeng Cao, Tong Zhou, Erlong Li, Ao Liu, Shengtao Zou, Chang Liu, Shuqi Mei, Elena Sizikova, Chao Zheng
Abstract
Nowadays, autonomous vehicle technology is becoming more and more mature. Critical to progress and safety, high-definition (HD) maps, a type of centimeter-level map collected using a laser sensor, provide accurate descriptions of the surrounding environment. The key challenge of HD map production is efficient, high-quality collection and annotation of large-volume datasets. Due to the demand for high quality, HD map production requires significant manual human effort to create annotations, a very time-consuming and costly process for the map industry. In order to reduce manual annotation burdens, many artificial intelligence (AI) algorithms have been developed to pre-label the HD maps. However, there still exists a large gap between AI algorithms and the traditional manual HD map production pipelines in accuracy and robustness. Furthermore, it is also very resource-costly to build large-scale annotated datasets and advanced machine learning algorithms for AI-based HD map automatic labeling systems. In this paper, we introduce the Tencent HD Map AI (THMA) system, an innovative end-to-end, AI-based, active learning HD map labeling system capable of producing and labeling HD maps with a scale of hundreds of thousands of kilometers. In THMA, we train AI models directly from massive HD map datasets via supervised, self-supervised, and weakly supervised learning to achieve high accuracy and efficiency required by downstream users. THMA has been deployed by the Tencent Map team to provide services to downstream companies and users, serving over 1,000 labeling workers and producing more than 30,000 kilometers of HD map data per day at most. More than 90 percent of the HD map data in Tencent Map is labeled automatically by THMA, accelerating the traditional HD map labeling process by more than ten times.
Similarity Contrastive Estimation for Image and Video Soft Contrastive Self-Supervised Learning
Authors: Julien Denize, Jaonary Rabarisoa, Astrid Orcesi, Romain Hérault
Abstract
Contrastive representation learning has proven to be an effective self-supervised learning method for images and videos. Most successful approaches are based on Noise Contrastive Estimation (NCE) and use different views of an instance as positives that should be contrasted with other instances, called negatives, that are considered as noise. However, several instances in a dataset are drawn from the same distribution and share underlying semantic information. A good data representation should contain relations between the instances, or semantic similarity and dissimilarity, that contrastive learning harms by considering all negatives as noise. To circumvent this issue, we propose a novel formulation of contrastive learning using semantic similarity between instances called Similarity Contrastive Estimation (SCE). Our training objective is a soft contrastive one that brings the positives closer and estimates a continuous distribution to push or pull negative instances based on their learned similarities. We validate empirically our approach on both image and video representation learning. We show that SCE performs competitively with the state of the art on the ImageNet linear evaluation protocol for fewer pretraining epochs and that it generalizes to several downstream image tasks. We also show that SCE reaches state-of-the-art results for pretraining video representation and that the learned representation can generalize to video downstream tasks.
Keyword: vision transformer
What Makes for Good Tokenizers in Vision Transformer?
Authors: Shengju Qian, Yi Zhu, Wenbo Li, Mu Li, Jiaya Jia
Subjects: Computer Vision and Pattern Recognition (cs.CV)
Abstract
The architecture of transformers, which recently witness booming applications in vision tasks, has pivoted against the widespread convolutional paradigm. Relying on the tokenization process that splits inputs into multiple tokens, transformers are capable of extracting their pairwise relationships using self-attention. While being the stemming building block of transformers, what makes for a good tokenizer has not been well understood in computer vision. In this work, we investigate this uncharted problem from an information trade-off perspective. In addition to unifying and understanding existing structural modifications, our derivation leads to better design strategies for vision tokenizers. The proposed Modulation across Tokens (MoTo) incorporates inter-token modeling capability through normalization. Furthermore, a regularization objective TokenProp is embraced in the standard training regime. Through extensive experiments on various transformer architectures, we observe both improved performance and intriguing properties of these two plug-and-play designs with negligible computational overhead. These observations further indicate the importance of the commonly-omitted designs of tokenizers in vision transformer.
Keyword: multimodal
MultiInstruct: Improving Multi-Modal Zero-Shot Learning via Instruction Tuning
Abstract
Instruction tuning, a new learning paradigm that fine-tunes pre-trained language models on tasks specified through instructions, has shown promising zero-shot performance on various natural language processing tasks. However, it's still not explored for vision and multimodal tasks. In this work, we introduce MultiInstruct, the first multimodal instruction tuning benchmark dataset that consists of 47 diverse multimodal tasks covering 11 broad categories. Each task is designed at least with 5,000 instances (input-out pairs) from existing open-source datasets and 5 expert-written instructions. We take OFA as the base pre-trained model for multimodal instruction tuning, and to improve its performance, we explore multiple transfer learning strategies to leverage the large-scale Natural Instructions dataset. Experimental results demonstrate its strong zero-shot performance on various unseen multimodal tasks and the benefit of transfer learning from text-only instructions. We also design a new evaluation metric: Sensitivity, to evaluate how sensitive the model is to the variety of instructions. Our results indicate that the model is less sensitive to the varying instructions after finetuning on a diverse set of tasks and instructions for each task.
Interactive Learning-from-Observation through multimodal human demonstration
Abstract
Learning-from-Observation (LfO) is a robot teaching framework for programming operations through few-shots human demonstration. While most previous LfO systems run with visual demonstration, recent research on robot teaching has shown the effectiveness of verbal instruction in making recognition robust and teaching interactive. To the best of our knowledge, however, few solutions have been proposed for LfO that utilizes verbal instruction, namely multimodal LfO. This paper aims to propose a practical pipeline for multimodal LfO. For input, an user temporally stops hand movements to match the granularity of human instructions with the granularity of robot execution. The pipeline recognizes tasks based on step-by-step verbal instructions accompanied by demonstrations. In addition, the recognition is made robust through interactions with the user. We test the pipeline on a real robot and show that the user can successfully teach multiple operations from multimodal demonstrations. The results suggest the utility of the proposed pipeline for multimodal LfO.
MAViC: Multimodal Active Learning for Video Captioning
Abstract
A large number of annotated video-caption pairs are required for training video captioning models, resulting in high annotation costs. Active learning can be instrumental in reducing these annotation requirements. However, active learning for video captioning is challenging because multiple semantically similar captions are valid for a video, resulting in high entropy outputs even for less-informative samples. Moreover, video captioning algorithms are multimodal in nature with a visual encoder and language decoder. Further, the sequential and combinatorial nature of the output makes the problem even more challenging. In this paper, we introduce MAViC which leverages our proposed Multimodal Semantics Aware Sequential Entropy (M-SASE) based acquisition function to address the challenges of active learning approaches for video captioning. Our approach integrates semantic similarity and uncertainty of both visual and language dimensions in the acquisition function. Our detailed experiments empirically demonstrate the efficacy of M-SASE for active learning for video captioning and improve on the baselines by a large margin.
OpenPack: A Large-scale Dataset for Recognizing Packaging Works in IoT-enabled Logistic Environments
Abstract
Unlike human daily activities, existing publicly available sensor datasets for work activity recognition in industrial domains are limited by difficulties in collecting realistic data as close collaboration with industrial sites is required. This also limits research on and development of AI methods for industrial applications. To address these challenges and contribute to research on machine recognition of work activities in industrial domains, in this study, we introduce a new large-scale dataset for packaging work recognition called OpenPack. OpenPack contains 53.8 hours of multimodal sensor data, including keypoints, depth images, acceleration data, and readings from IoT-enabled devices (e.g., handheld barcode scanners used in work procedures), collected from 16 distinct subjects with different levels of packaging work experience. On the basis of this dataset, we propose a neural network model designed to recognize work activities, which efficiently fuses sensor data and readings from IoT-enabled devices by processing them within different streams in a ladder-shaped architecture, and the experiment showed the effectiveness of the architecture. We believe that OpenPack will contribute to the community of action/activity recognition with sensors. OpenPack dataset is available at https://open-pack.github.io/.
Keyword: CLIP
Contrastive Language-Vision AI Models Pretrained on Web-Scraped Multimodal Data Exhibit Sexual Objectification Bias
Authors: Robert Wolfe, Yiwei Yang, Bill Howe, Aylin Caliskan
Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Abstract
Nine language-vision AI models trained on web scrapes with the Contrastive Language-Image Pretraining (CLIP) objective are evaluated for evidence of a bias studied by psychologists: the sexual objectification of girls and women, which occurs when a person's human characteristics are disregarded and the person is treated as a body or a collection of body parts. A first experiment uses standardized images of women from the Sexual OBjectification and EMotion Database, and finds that, commensurate with prior research in psychology, human characteristics are disassociated from images of objectified women: the model's recognition of emotional state is mediated by whether the subject is fully or partially clothed. Embedding association tests (EATs) return significant effect sizes for both anger (d >.8) and sadness (d >.5). A second experiment measures the effect in a representative application: an automatic image captioner (Antarctic Captions) includes words denoting emotion less than 50% as often for images of partially clothed women than for images of fully clothed women. A third experiment finds that images of female professionals (scientists, doctors, executives) are likely to be associated with sexual descriptions relative to images of male professionals. A fourth experiment shows that a prompt of "a [age] year old girl" generates sexualized images (as determined by an NSFW classifier) up to 73% of the time for VQGAN-CLIP (age 17), and up to 40% of the time for Stable Diffusion (ages 14 and 18); the corresponding rate for boys never surpasses 9%. The evidence indicates that language-vision AI models trained on automatically collected web scrapes learn biases of sexual objectification, which propagate to downstream applications.
3D Highlighter: Localizing Regions on 3D Shapes via Text Descriptions
Authors: Dale Decatur, Itai Lang, Rana Hanocka
Subjects: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV)
Abstract
We present 3D Highlighter, a technique for localizing semantic regions on a mesh using text as input. A key feature of our system is the ability to interpret "out-of-domain" localizations. Our system demonstrates the ability to reason about where to place non-obviously related concepts on an input 3D shape, such as adding clothing to a bare 3D animal model. Our method contextualizes the text description using a neural field and colors the corresponding region of the shape using a probability-weighted blend. Our neural optimization is guided by a pre-trained CLIP encoder, which bypasses the need for any 3D datasets or 3D annotations. Thus, 3D Highlighter is highly flexible, general, and capable of producing localizations on a myriad of input shapes. Our code is publicly available at https://github.com/threedle/3DHighlighter.
Keyword: metric learning
There is no result
Keyword: image retrieval
There is no result
Keyword: self-supervised
MolCPT: Molecule Continuous Prompt Tuning to Generalize Molecular Representation Learning
Continual Contrastive Finetuning Improves Low-Resource Relation Extraction
Joint Embedding of 2D and 3D Networks for Medical Image Anomaly Detection
TruFor: Leveraging all-round clues for trustworthy image forgery detection and localization
THMA: Tencent HD Map AI System for Creating HD Map Annotations
Similarity Contrastive Estimation for Image and Video Soft Contrastive Self-Supervised Learning
Keyword: vision transformer
What Makes for Good Tokenizers in Vision Transformer?
Keyword: multimodal
MultiInstruct: Improving Multi-Modal Zero-Shot Learning via Instruction Tuning
Interactive Learning-from-Observation through multimodal human demonstration
MAViC: Multimodal Active Learning for Video Captioning
OpenPack: A Large-scale Dataset for Recognizing Packaging Works in IoT-enabled Logistic Environments
Keyword: CLIP
Contrastive Language-Vision AI Models Pretrained on Web-Scraped Multimodal Data Exhibit Sexual Objectification Bias
3D Highlighter: Localizing Regions on 3D Shapes via Text Descriptions
Keyword: DALLE
There is no result