Auscultation is crucial for diagnosing lung diseases. The COVID-19 pandemic has revealed the limitations of traditional, in-person lung sound assessments. To overcome these issues, advancements in digital stethoscopes and artificial intelligence (AI) have led to the development of new diagnostic methods. In this context, our study aims to use smartphone microphones to record and analyze lung sounds. We faced two major challenges: the difference in audio style between electronic stethoscopes and smartphone microphones, and the variability among patients. To address these challenges, we developed a method called Patient Domain Supervised Contrastive Learning (PD-SCL). By integrating this method with the Audio Spectrogram Transformer (AST) model, we significantly improved its performance by 2.4\% compared to the original AST model. This progress demonstrates that smartphones can effectively diagnose lung sounds, addressing inconsistencies in patient data and showing potential for broad use beyond traditional clinical settings. Our research contributes to making lung disease detection more accessible in the post-COVID-19 world.
Effective video tokenization is critical for scaling transformer models for long videos. Current approaches tokenize videos using space-time patches, leading to excessive tokens and computational inefficiencies. The best token reduction strategies degrade performance and barely reduce the number of tokens when the camera moves. We introduce grounded video tokenization, a paradigm that organizes tokens based on panoptic sub-object trajectories rather than fixed patches. Our method aligns with fundamental perceptual principles, ensuring that tokenization reflects scene complexity rather than video duration. We propose TrajViT, a video encoder that extracts object trajectories and converts them into semantically meaningful tokens, significantly reducing redundancy while maintaining temporal coherence. Trained with contrastive learning, TrajViT significantly outperforms space-time ViT (ViT3D) across multiple video understanding benchmarks, e.g., TrajViT outperforms ViT3D by a large margin of 6% top-5 recall in average at video-text retrieval task with 10x token deduction. We also show TrajViT as a stronger model than ViT3D for being the video encoder for modern VideoLLM, obtaining an average of 5.2% performance improvement across 6 VideoQA benchmarks while having 4x faster training time and 18x less inference FLOPs. TrajViT is the first efficient encoder to consistently outperform ViT3D across diverse video analysis tasks, making it a robust and scalable solution.
Recent works of music representation learning mainly focus on learning acoustic music representations with unlabeled audios or further attempt to acquire multi-modal music representations with scarce annotated audio-text pairs. They either ignore the language semantics or rely on labeled audio datasets that are difficult and expensive to create. Moreover, merely modeling semantic space usually fails to achieve satisfactory performance on music recommendation tasks since the user preference space is ignored. In this paper, we propose a novel Hierarchical Two-stage Contrastive Learning (HTCL) method that models similarity from the semantic perspective to the user perspective hierarchically to learn a comprehensive music representation bridging the gap between semantic and user preference spaces. We devise a scalable audio encoder and leverage a pre-trained BERT model as the text encoder to learn audio-text semantics via large-scale contrastive pre-training. Further, we explore a simple yet effective way to exploit interaction data from our online music platform to adapt the semantic space to user preference space via contrastive fine-tuning, which differs from previous works that follow the idea of collaborative filtering. As a result, we obtain a powerful audio encoder that not only distills language semantics from the text encoder but also models similarity in user preference space with the integrity of semantic space preserved. Experimental results on both music semantic and recommendation tasks confirm the effectiveness of our method.
Graph representation learning is effective for obtaining a meaningful latent space utilizing the structure of graph data and is widely applied, including biological networks. In particular, Graph Contrastive Learning (GCL) has emerged as a powerful self-supervised method that relies on applying perturbations to graphs for data augmentation. However, when applying existing GCL methods to biological networks such as Gene Regulatory Networks (GRNs), they overlooked meaningful biologically relevant perturbations, e.g., gene knockdowns. In this study, we introduce SupGCL (Supervised Graph Contrastive Learning), a novel GCL method for GRNs that directly incorporates biological perturbations derived from gene knockdown experiments as the supervision. SupGCL mathematically extends existing GCL methods that utilize non-biological perturbations to probabilistic models that introduce actual biological gene perturbation utilizing gene knockdown data. Using the GRN representation obtained by our proposed method, our aim is to improve the performance of biological downstream tasks such as patient hazard prediction and disease subtype classification (graph-level task), and gene function classification (node-level task). We applied SupGCL on real GRN datasets derived from patients with multiple types of cancer, and in all experiments SupGCL achieves better performance than state-of-the-art baselines.
Student engagement plays a crucial role in the successful delivery of educational programs. Automated engagement measurement helps instructors monitor student participation, identify disengagement, and adapt their teaching strategies to enhance learning outcomes effectively. This paper identifies two key challenges in this problem: class imbalance and incorporating order into engagement levels rather than treating it as mere categories. Then, a novel approach to video-based student engagement measurement in virtual learning environments is proposed that utilizes supervised contrastive learning for ordinal classification of engagement. Various affective and behavioral features are extracted from video samples and utilized to train ordinal classifiers within a supervised contrastive learning framework (with a sequential classifier as the encoder). A key step involves the application of diverse time-series data augmentation techniques to these feature vectors, enhancing model training. The effectiveness of the proposed method was evaluated using a publicly available dataset for engagement measurement, DAiSEE, containing videos of students who participated in virtual learning programs. The results demonstrate the robust ability of the proposed method for the classification of the engagement level. This approach promises a significant contribution to understanding and enhancing student engagement in virtual learning environments.
While significant progress has been made with dual- and bi-encoder dense retrievers, they often struggle on queries with logical connectives, a use case that is often overlooked yet important in downstream applications. Current dense retrievers struggle with such queries, such that the retrieved results do not respect the logical constraints implied in the queries. To address this challenge, we introduce LogiCoL, a logically-informed contrastive learning objective for dense retrievers. LogiCoL builds upon in-batch supervised contrastive learning, and learns dense retrievers to respect the subset and mutually-exclusive set relation between query results via two sets of soft constraints expressed via t-norm in the learning objective. We evaluate the effectiveness of LogiCoL on the task of entity retrieval, where the model is expected to retrieve a set of entities in Wikipedia that satisfy the implicit logical constraints in the query. We show that models trained with LogiCoL yield improvement both in terms of retrieval performance and logical consistency in the results. We provide detailed analysis and insights to uncover why queries with logical connectives are challenging for dense retrievers and why LogiCoL is most effective.
Controllable code generation, the ability to synthesize code that follows a specified style while maintaining functionality, remains a challenging task. We propose a two-stage training framework combining contrastive learning and conditional decoding to enable flexible style control. The first stage aligns code style representations with semantic and structural features. In the second stage, we fine-tune a language model (e.g., Flan-T5) conditioned on the learned style vector to guide generation. Our method supports style interpolation and user personalization via lightweight mixing. Compared to prior work, our unified framework offers improved stylistic control without sacrificing code correctness. This is among the first approaches to combine contrastive alignment with conditional decoding for style-guided code generation.
We investigate improving the retrieval effectiveness of embedding models through the lens of corpus-specific fine-tuning. Prior work has shown that fine-tuning with queries generated using a dataset's retrieval corpus can boost retrieval effectiveness for the dataset. However, we find that surprisingly, fine-tuning using the conventional InfoNCE contrastive loss often reduces effectiveness in state-of-the-art models. To overcome this, we revisit cross-encoder listwise distillation and demonstrate that, unlike using contrastive learning alone, listwise distillation can help more consistently improve retrieval effectiveness across multiple datasets. Additionally, we show that synthesizing more training data using diverse query types (such as claims, keywords, and questions) yields greater effectiveness than using any single query type alone, regardless of the query type used in evaluation. Our findings further indicate that synthetic queries offer comparable utility to human-written queries for training. We use our approach to train an embedding model that achieves state-of-the-art effectiveness among BERT embedding models. We release our model and both query generation and training code to facilitate further research.
Graph Contrastive Learning (GCL), which fuses graph neural networks with contrastive learning, has evolved as a pivotal tool in user-item recommendations. While promising, existing GCL methods often lack explicit modeling of hierarchical item structures, which represent item similarities across varying resolutions. Such hierarchical item structures are ubiquitous in various items (e.g., online products and local businesses), and reflect their inherent organizational properties that serve as critical signals for enhancing recommendation accuracy. In this paper, we propose Hierarchical Graph Contrastive Learning (HGCL), a novel GCL method that incorporates hierarchical item structures for user-item recommendations. First, HGCL pre-trains a GCL module using cross-layer contrastive learning to obtain user and item representations. Second, HGCL employs a representation compression and clustering method to construct a two-hierarchy user-item bipartite graph. Ultimately, HGCL fine-tunes user and item representations by learning on the hierarchical graph, and then provides recommendations based on user-item interaction scores. Experiments on three widely adopted benchmark datasets ranging from 70K to 382K nodes confirm the superior performance of HGCL over existing baseline models, highlighting the contribution of hierarchical item structures in enhancing GCL methods for recommendation tasks.
Accurate predictions using sequential spatiotemporal data are crucial for various applications. Utilizing real-world data, we aim to learn the intent of a smart device user within confined areas of a vehicle's surroundings. However, in real-world scenarios, environmental factors and sensor limitations result in non-stationary and irregularly sampled data, posing significant challenges. To address these issues, we developed a Transformer-based approach, STaRFormer, which serves as a universal framework for sequential modeling. STaRFormer employs a novel, dynamic attention-based regional masking scheme combined with semi-supervised contrastive learning to enhance task-specific latent representations. Comprehensive experiments on 15 datasets varying in types (including non-stationary and irregularly sampled), domains, sequence lengths, training samples, and applications, demonstrate the efficacy and practicality of STaRFormer. We achieve notable improvements over state-of-the-art approaches. Code and data will be made available.
Current debiasing approaches often result a degradation in model capabilities such as factual accuracy and knowledge retention. Through systematic evaluation across multiple benchmarks, we demonstrate that existing debiasing methods face fundamental trade-offs, particularly in smaller models, leading to reduced truthfulness, knowledge loss, or unintelligible outputs. To address these limitations, we propose a contrastive learning framework that learns through carefully constructed positive and negative examples. Our approach introduces contrast computation and dynamic loss scaling to balance bias mitigation with faithfulness preservation. Experimental results across multiple model scales demonstrate that our method achieves substantial improvements in both toxicity reduction and faithfulness preservation. Most importantly, we show that our framework is the first to consistently improve both metrics simultaneously, avoiding the capability degradation characteristic of existing approaches. These results suggest that explicit modeling of both positive and negative examples through contrastive learning could be a promising direction for reducing the alignment tax in language model debiasing.
Although graph contrastive learning (GCL) has been widely investigated, it is still a challenge to generate effective and stable graph augmentations. Existing methods often apply heuristic augmentation like random edge dropping, which may disrupt important graph structures and result in unstable GCL performance. In this paper, we propose Positive-incentive Noise driven Graph Data Augmentation (PiNGDA), where positive-incentive noise (pi-noise) scientifically analyzes the beneficial effect of noise under the information theory. To bridge the standard GCL and pi-noise framework, we design a Gaussian auxiliary variable to convert the loss function to information entropy. We prove that the standard GCL with pre-defined augmentations is equivalent to estimate the beneficial noise via the point estimation. Following our analysis, PiNGDA is derived from learning the beneficial noise on both topology and attributes through a trainable noise generator for graph augmentations, instead of the simple estimation. Since the generator learns how to produce beneficial perturbations on graph topology and node attributes, PiNGDA is more reliable compared with the existing methods. Extensive experimental results validate the effectiveness and stability of PiNGDA.
With the advancement of edge computing, federated learning (FL) displays a bright promise as a privacy-preserving collaborative learning paradigm. However, one major challenge for FL is the data heterogeneity issue, which refers to the biased labeling preferences among multiple clients, negatively impacting convergence and model performance. Most previous FL methods attempt to tackle the data heterogeneity issue locally or globally, neglecting underlying class-wise structure information contained in each client. In this paper, we first study how data heterogeneity affects the divergence of the model and decompose it into local, global, and sampling drift sub-problems. To explore the potential of using intra-client class-wise structural knowledge in handling these drifts, we thus propose Federated Learning with Structural Knowledge Collaboration (FedSKC). The key idea of FedSKC is to extract and transfer domain preferences from inter-client data distributions, offering diverse class-relevant knowledge and a fair convergent signal. FedSKC comprises three components: i) local contrastive learning, to prevent weight divergence resulting from local training; ii) global discrepancy aggregation, which addresses the parameter deviation between the server and clients; iii) global period review, correcting for the sampling drift introduced by the server randomly selecting devices. We have theoretically analyzed FedSKC under non-convex objectives and empirically validated its superiority through extensive experimental results.
With the increasing availability of geospatial datasets, researchers have explored region representation learning (RRL) to analyze complex region characteristics. Recent RRL methods use contrastive learning (CL) to capture shared information between two modalities but often overlook task-relevant unique information specific to each modality. Such modality-specific details can explain region characteristics that shared information alone cannot capture. Bringing information factorization to RRL can address this by factorizing multimodal data into shared and unique information. However, existing factorization approaches focus on two modalities, whereas RRL can benefit from various geospatial data. Extending factorization beyond two modalities is non-trivial because modeling high-order relationships introduces a combinatorial number of learning objectives, increasing model complexity. We introduce Cross modal Knowledge Injected Embedding, an information factorization approach for RRL that captures both shared and unique representations. CooKIE uses a pairwise inter-view learning approach that captures high-order information without modeling high-order dependency, avoiding exhaustive combinations. We evaluate CooKIE on three regression tasks and a land use classification task in New York City and Delhi, India. Results show that CooKIE outperforms existing RRL methods and a factorized RRL model, capturing multimodal information with fewer training parameters and floating-point operations per second (FLOPs). We release the code: https://github.com/MinNamgung/CooKIE.
Detecting Mild Cognitive Impairment from picture descriptions is critical yet challenging, especially in multilingual and multiple picture settings. Prior work has primarily focused on English speakers describing a single picture (e.g., the 'Cookie Theft'). The TAUKDIAL-2024 challenge expands this scope by introducing multilingual speakers and multiple pictures, which presents new challenges in analyzing picture-dependent content. To address these challenges, we propose a framework with three components: (1) enhancing discriminative representation learning via supervised contrastive learning, (2) involving image modality rather than relying solely on speech and text modalities, and (3) applying a Product of Experts (PoE) strategy to mitigate spurious correlations and overfitting. Our framework improves MCI detection performance, achieving a +7.1% increase in Unweighted Average Recall (UAR) (from 68.1% to 75.2%) and a +2.9% increase in F1 score (from 80.6% to 83.5%) compared to the text unimodal baseline. Notably, the contrastive learning component yields greater gains for the text modality compared to speech. These results highlight our framework's effectiveness in multilingual and multi-picture MCI detection.
Benefiting from contrastively trained visual encoders on large-scale natural scene images, Large Multimodal Models (LMMs) have achieved remarkable performance across various visual perception tasks. However, the inherent limitations of contrastive learning upon summarized descriptions fundamentally restrict the capabilities of models in meticulous reasoning, particularly in crucial scenarios of geometric problem-solving. To enhance geometric understanding, we propose a novel hard negative contrastive learning framework for the vision encoder, which combines image-based contrastive learning using generation-based hard negatives created by perturbing diagram generation code, and text-based contrastive learning using rule-based negatives derived from modified geometric descriptions and retrieval-based negatives selected based on caption similarity. We train CLIP using our strong negative learning method, namely MMCLIP (Multimodal Math CLIP), and subsequently train an LMM for geometric problem-solving. Experiments show that our trained model, MMGeoLM, significantly outperforms other open-source models on three geometric reasoning benchmarks. Even with a size of 7B, it can rival powerful closed-source models like GPT-4o. We further study the impact of different negative sample construction methods and the number of negative samples on the geometric reasoning performance of LMM, yielding fruitful conclusions. The code and dataset are available at https://github.com/THU-KEG/MMGeoLM.
Zero-Shot Composed Image Retrieval (ZS-CIR) aims to retrieve target images given a compositional query, consisting of a reference image and a modifying text-without relying on annotated training data. Existing approaches often generate a synthetic target text using large language models (LLMs) to serve as an intermediate anchor between the compositional query and the target image. Models are then trained to align the compositional query with the generated text, and separately align images with their corresponding texts using contrastive learning. However, this reliance on intermediate text introduces error propagation, as inaccuracies in query-to-text and text-to-image mappings accumulate, ultimately degrading retrieval performance. To address these problems, we propose a novel framework by employing a Multimodal Reasoning Agent (MRA) for ZS-CIR. MRA eliminates the dependence on textual intermediaries by directly constructing triplets, <reference image, modification text, target image>, using only unlabeled image data. By training on these synthetic triplets, our model learns to capture the relationships between compositional queries and candidate images directly. Extensive experiments on three standard CIR benchmarks demonstrate the effectiveness of our approach. On the FashionIQ dataset, our method improves Average R@10 by at least 7.5\% over existing baselines; on CIRR, it boosts R@1 by 9.6\%; and on CIRCO, it increases mAP@5 by 9.5\%.
The diagram is a visual representation of a relationship illustrated with edges (lines or arrows), which is widely used in industrial and scientific communication. Although recognizing diagrams is essential for vision language models (VLMs) to comprehend domain-specific knowledge, recent studies reveal that many VLMs fail to identify edges in images. We hypothesize that these failures stem from an over-reliance on textual and positional biases, preventing VLMs from learning explicit edge features. Based on this idea, we empirically investigate whether the image encoder in VLMs can learn edge representation through training on a diagram dataset in which edges are biased neither by textual nor positional information. To this end, we conduct contrastive learning on an artificially generated diagram--caption dataset to train an image encoder and evaluate its diagram-related features on three tasks: probing, image retrieval, and captioning. Our results show that the finetuned model outperforms pretrained CLIP in all tasks and surpasses zero-shot GPT-4o and LLaVA-Mistral in the captioning task. These findings confirm that eliminating textual and positional biases fosters accurate edge recognition in VLMs, offering a promising path for advancing diagram understanding.
We introduce FruitNeRF++, a novel fruit-counting approach that combines contrastive learning with neural radiance fields to count fruits from unstructured input photographs of orchards. Our work is based on FruitNeRF, which employs a neural semantic field combined with a fruit-specific clustering approach. The requirement for adaptation for each fruit type limits the applicability of the method, and makes it difficult to use in practice. To lift this limitation, we design a shape-agnostic multi-fruit counting framework, that complements the RGB and semantic data with instance masks predicted by a vision foundation model. The masks are used to encode the identity of each fruit as instance embeddings into a neural instance field. By volumetrically sampling the neural fields, we extract a point cloud embedded with the instance features, which can be clustered in a fruit-agnostic manner to obtain the fruit count. We evaluate our approach using a synthetic dataset containing apples, plums, lemons, pears, peaches, and mangoes, as well as a real-world benchmark apple dataset. Our results demonstrate that FruitNeRF++ is easier to control and compares favorably to other state-of-the-art methods.
Self-Supervised Learning (SSL) enables us to pre-train foundation models without costly labeled data. Among SSL methods, Contrastive Learning (CL) methods are better at obtaining accurate semantic representations in noise interference. However, due to the significant domain gap, while CL methods have achieved great success in many computer vision tasks, they still require specific adaptation for Remote Sensing (RS) images. To this end, we present a novel self-supervised method called PerA, which produces all-purpose RS features through semantically Perfectly Aligned sample pairs. Specifically, PerA obtains features from sampled views by applying spatially disjoint masks to augmented images rather than random cropping. Our framework provides high-quality features by ensuring consistency between teacher and student and predicting learnable mask tokens. Compared to previous contrastive methods, our method demonstrates higher memory efficiency and can be trained with larger batches due to its sparse inputs. Additionally, the proposed method demonstrates remarkable adaptability to uncurated RS data and reduce the impact of the potential semantic inconsistency. We also collect an unlabeled pre-training dataset, which contains about 5 million RS images. We conducted experiments on multiple downstream task datasets and achieved performance comparable to previous state-of-the-art methods with a limited model scale, demonstrating the effectiveness of our approach. We hope this work will contribute to practical remote sensing interpretation works.