Contrastive language-audio pre-training (CLAP) has addressed audio-language tasks such as audio-text retrieval by aligning audio and text in a common feature space. While CLAP addresses general audio-language tasks, its audio features do not generalize well in audio tasks. In contrast, self-supervised learning (SSL) models learn general-purpose audio features that perform well in diverse audio tasks. We pursue representation learning that can be widely used in audio applications and hypothesize that a method that learns both general audio features and CLAP features should achieve our goal, which we call a general-purpose audio-language representation. To implement our hypothesis, we propose M2D2, a second-generation masked modeling duo (M2D) that combines an SSL M2D and CLAP. M2D2 learns two types of features using two modalities (audio and text) in a two-stage training process. It also utilizes advanced LLM-based sentence embeddings in CLAP training for powerful semantic supervision. In the first stage, M2D2 learns generalizable audio features from M2D and CLAP, where CLAP aligns the features with the fine LLM-based semantic embeddings. In the second stage, it learns CLAP features using the audio features learned from the LLM-based embeddings. Through these pre-training stages, M2D2 should enhance generalizability and performance in its audio and CLAP features. Experiments validated that M2D2 achieves effective general-purpose audio-language representation, highlighted with SOTA fine-tuning mAP of 49.0 for AudioSet, SOTA performance in music tasks, and top-level performance in audio-language tasks.
Due to limitations such as geographic, physical, or economic factors, collected seismic data often have missing traces. Traditional seismic data reconstruction methods face the challenge of selecting numerous empirical parameters and struggle to handle large-scale continuous missing traces. With the advancement of deep learning, various diffusion models have demonstrated strong reconstruction capabilities. However, these UNet-based diffusion models require significant computational resources and struggle to learn the correlation between different traces in seismic data. To address the complex and irregular missing situations in seismic data, we propose a latent diffusion transformer utilizing representation learning for seismic data reconstruction. By employing a mask modeling scheme based on representation learning, the representation module uses the token sequence of known data to infer the token sequence of unknown data, enabling the reconstructed data from the diffusion model to have a more consistent data distribution and better correlation and accuracy with the known data. We propose the Representation Diffusion Transformer architecture, and a relative positional bias is added when calculating attention, enabling the diffusion model to achieve global modeling capability for seismic data. Using a pre-trained data compression model compresses the training and inference processes of the diffusion model into a latent space, which, compared to other diffusion model-based reconstruction methods, reduces computational and inference costs. Reconstruction experiments on field and synthetic datasets indicate that our method achieves higher reconstruction accuracy than existing methods and can handle various complex missing scenarios.
Existing foundation models, such as CLIP, aim to learn a unified embedding space for multimodal data, enabling a wide range of downstream web-based applications like search, recommendation, and content classification. However, these models often overlook the inherent graph structures in multimodal datasets, where entities and their relationships are crucial. Multimodal graphs (MMGs) represent such graphs where each node is associated with features from different modalities, while the edges capture the relationships between these entities. On the other hand, existing graph foundation models primarily focus on text-attributed graphs (TAGs) and are not designed to handle the complexities of MMGs. To address these limitations, we propose UniGraph2, a novel cross-domain graph foundation model that enables general representation learning on MMGs, providing a unified embedding space. UniGraph2 employs modality-specific encoders alongside a graph neural network (GNN) to learn a unified low-dimensional embedding space that captures both the multimodal information and the underlying graph structure. We propose a new cross-domain multi-graph pre-training algorithm at scale to ensure effective transfer learning across diverse graph domains and modalities. Additionally, we adopt a Mixture of Experts (MoE) component to align features from different domains and modalities, ensuring coherent and robust embeddings that unify the information across modalities. Extensive experiments on a variety of multimodal graph tasks demonstrate that UniGraph2 significantly outperforms state-of-the-art models in tasks such as representation learning, transfer learning, and multimodal generative tasks, offering a scalable and flexible solution for learning on MMGs.
Quantum Graph Neural Networks (QGNNs) present a promising approach for combining quantum computing with graph-structured data processing. While classical Graph Neural Networks (GNNs) are renowned for their scalability and robustness, existing QGNNs often lack flexibility due to graph-specific quantum circuit designs, limiting their applicability to a narrower range of graph-structured problems, falling short of real-world scenarios. To address these limitations, we propose a versatile QGNN framework inspired by the classical GraphSAGE approach, utilizing quantum models as aggregators. In this work, we integrate established techniques for inductive representation learning on graphs with parametrized quantum convolutional and pooling layers, effectively bridging classical and quantum paradigms. The convolutional layer is flexible, enabling tailored designs for specific problems. Benchmarked on a node regression task with the QM9 dataset, we demonstrate that our framework successfully models a non-trivial molecular dataset, achieving performance comparable to classical GNNs. In particular, we show that our quantum approach exhibits robust generalization across molecules with varying numbers of atoms without requiring circuit modifications, slightly outperforming classical GNNs. Furthermore, we numerically investigate the scalability of the QGNN framework. Specifically, we demonstrate the absence of barren plateaus in our architecture as the number of qubits increases, suggesting that the proposed quantum model can be extended to handle larger and more complex graph-based problems effectively.
Graph layouts and node embeddings are two distinct paradigms for non-parametric graph representation learning. In the former, nodes are embedded into 2D space for visualization purposes. In the latter, nodes are embedded into a high-dimensional vector space for downstream processing. State-of-the-art algorithms for these two paradigms, force-directed layouts and random-walk-based contrastive learning (such as DeepWalk and node2vec), have little in common. In this work, we show that both paradigms can be approached with a single coherent framework based on established neighbor embedding methods. Specifically, we introduce graph t-SNE, a neighbor embedding method for two-dimensional graph layouts, and graph CNE, a contrastive neighbor embedding method that produces high-dimensional node representations by optimizing the InfoNCE objective. We show that both graph t-SNE and graph CNE strongly outperform state-of-the-art algorithms in terms of local structure preservation, while being conceptually simpler.
Understanding the representation shift on Vision Language Models like CLIP under different augmentations provides valuable insights on Mechanistic Interpretability. In this study, we show the shift on CLIP's embeddings on 9 common augmentation techniques: noise, blur, color jitter, scale and rotate, flip, elastic and perspective transforms, random brightness and contrast, and coarse dropout of pixel blocks. We scrutinize the embedding shifts under similarity on attention map, patch, edge, detail preservation, cosine similarity, L2 distance, pairwise distance and dendrogram clusters and provide qualitative analysis on sample images. Our findings suggest certain augmentations like noise, perspective transform and shift scaling have higher degree of drastic impact on embedding shift. This study provides a concrete foundation for future work on VLM's robustness for mechanical interpretation and adversarial data defense.
Contrastive Language-Image Pretraining (CLIP) has achieved remarkable success in cross-modal tasks such as zero-shot image classification and text-image retrieval by effectively aligning visual and textual representations. However, the theoretical foundations underlying CLIP's strong generalization remain unclear. In this work, we address this gap by proposing the Cross-modal Information Bottleneck (CIB) framework. CIB offers a principled interpretation of CLIP's contrastive learning objective as an implicit Information Bottleneck optimization. Under this view, the model maximizes shared cross-modal information while discarding modality-specific redundancies, thereby preserving essential semantic alignment across modalities. Building on this insight, we introduce a Cross-modal Information Bottleneck Regularization (CIBR) method that explicitly enforces these IB principles during training. CIBR introduces a penalty term to discourage modality-specific redundancy, thereby enhancing semantic alignment between image and text features. We validate CIBR on extensive vision-language benchmarks, including zero-shot classification across seven diverse image datasets and text-image retrieval on MSCOCO and Flickr30K. The results show consistent performance gains over standard CLIP. These findings provide the first theoretical understanding of CLIP's generalization through the IB lens. They also demonstrate practical improvements, offering guidance for future cross-modal representation learning.
Motivation: In recent years, protein function prediction has broken through the bottleneck of sequence features, significantly improving prediction accuracy using high-precision protein structures predicted by AlphaFold2. While single-species protein function prediction methods have achieved remarkable success, multi-species protein function prediction methods are still in the stage of using PPI networks and sequence features. Providing effective cross-species label propagation for species with sparse protein annotations remains a challenging issue. To address this problem, we propose the MSNGO model, which integrates structural features and network propagation methods. Our validation shows that using structural features can significantly improve the accuracy of multi-species protein function prediction. Results: We employ graph representation learning techniques to extract amino acid representations from protein structure contact maps and train a structural model using a graph convolution pooling module to derive protein-level structural features. After incorporating the sequence features from ESM-2, we apply a network propagation algorithm to aggregate information and update node representations within a heterogeneous network. The results demonstrate that MSNGO outperforms previous multi-species protein function prediction methods that rely on sequence features and PPI networks. Availability: https://github.com/blingbell/MSNGO.
The semi-airborne transient electromagnetic method (SATEM) is capable of conducting rapid surveys over large-scale and hard-to-reach areas. However, the acquired signals are often contaminated by complex noise, which can compromise the accuracy of subsequent inversion interpretations. Traditional denoising techniques primarily rely on parameter selection strategies, which are insufficient for processing field data in noisy environments. With the advent of deep learning, various neural networks have been employed for SATEM signal denoising. However, existing deep learning methods typically use single-mapping learning approaches that struggle to effectively separate signal from noise. These methods capture only partial information and lack interpretability. To overcome these limitations, we propose an interpretable decoupled representation learning framework, termed DREMnet, that disentangles data into content and context factors, enabling robust and interpretable denoising in complex conditions. To address the limitations of CNN and Transformer architectures, we utilize the RWKV architecture for data processing and introduce the Contextual-WKV mechanism, which allows unidirectional WKV to perform bidirectional signal modeling. Our proposed Covering Embedding technique retains the strong local perception of convolutional networks through stacked embedding. Experimental results on test datasets demonstrate that the DREMnet method outperforms existing techniques, with processed field data that more accurately reflects the theoretical signal, offering improved identification of subsurface electrical structures.
The extraction of geoelectric structural information from airborne transient electromagnetic(ATEM)data primarily involves data processing and inversion. Conventional methods rely on empirical parameter selection, making it difficult to process complex field data with high noise levels. Additionally, inversion computations are time consuming and often suffer from multiple local minima. Existing deep learning-based approaches separate the data processing steps, where independently trained denoising networks struggle to ensure the reliability of subsequent inversions. Moreover, end to end networks lack interpretability. To address these issues, we propose a unified and interpretable deep learning inversion paradigm based on disentangled representation learning. The network explicitly decomposes noisy data into noise and signal factors, completing the entire data processing workflow based on the signal factors while incorporating physical information for guidance. This approach enhances the network's reliability and interpretability. The inversion results on field data demonstrate that our method can directly use noisy data to accurately reconstruct the subsurface electrical structure. Furthermore, it effectively processes data severely affected by environmental noise, which traditional methods struggle with, yielding improved lateral structural resolution.
Unsupervised representation learning has been widely explored across various modalities, including neural architectures, where it plays a key role in downstream applications like Neural Architecture Search (NAS). These methods typically learn an unsupervised representation space before generating/ sampling architectures for the downstream search. A common approach involves the use of Variational Autoencoders (VAEs) to map discrete architectures onto a continuous representation space, however, sampling from these spaces often leads to a high percentage of invalid or duplicate neural architectures. This could be due to the unnatural mapping of inherently discrete architectural space onto a continuous space, which emphasizes the need for a robust discrete representation of these architectures. To address this, we introduce a Vector Quantized Variational Autoencoder (VQ-VAE) to learn a discrete latent space more naturally aligned with the discrete neural architectures. In contrast to VAEs, VQ-VAEs (i) map each architecture into a discrete code sequence and (ii) allow the prior to be learned by any generative model rather than assuming a normal distribution. We then represent these architecture latent codes as numerical sequences and train a text-to-text model leveraging a Large Language Model to learn and generate sequences representing architectures. We experiment our method with Inception/ ResNet-like cell-based search spaces, namely NAS-Bench-101 and NAS-Bench-201. Compared to VAE-based methods, our approach improves the generation of valid and unique architectures by over 80% on NASBench-101 and over 8% on NASBench-201. Finally, we demonstrate the applicability of our method in NAS employing a sequence-modeling-based NAS algorithm.
The rapid growth of unlabeled time series data, driven by the Internet of Things (IoT), poses significant challenges in uncovering underlying patterns. Traditional unsupervised clustering methods often fail to capture the complex nature of time series data. Recent deep learning-based clustering approaches, while effective, struggle with insufficient representation learning and the integration of clustering objectives. To address these issues, we propose a fuzzy cluster-aware contrastive clustering framework (FCACC) that jointly optimizes representation learning and clustering. Our approach introduces a novel three-view data augmentation strategy to enhance feature extraction by leveraging various characteristics of time series data. Additionally, we propose a cluster-aware hard negative sample generation mechanism that dynamically constructs high-quality negative samples using clustering structure information, thereby improving the model's discriminative ability. By leveraging fuzzy clustering, FCACC dynamically generates cluster structures to guide the contrastive learning process, resulting in more accurate clustering. Extensive experiments on 40 benchmark datasets show that FCACC outperforms the selected baseline methods (eight in total), providing an effective solution for unsupervised time series learning.
Over the past ten years, graph representation learning has garnered a lot of attention due to the variety of graph-structured data and its efficiency in both time and space. One essential method for obtaining effective graph representations is graph pooling. Numerous studies on the graph pooling technique have been conducted. Cutting-edge results on a range of graph representation learning tasks were made possible by the combination of graph neural networks and self-attention mechanisms. Nevertheless, the attention mechanism has limitations since it ignores nodes that have no direct connection via an edge but provide valuable network context information. This paper proposes a graph pooling approach based on Personalized PageRank and self-attention, which improves the model to take into account both node properties and graph structure. The experimental findings indicate that, with a suitable number of parameters, the MAGPool approach delivers greater accuracy on the benchmark datasets.
In the realm of stock prediction, machine learning models encounter considerable obstacles due to the inherent low signal-to-noise ratio and the nonstationary nature of financial markets. These challenges often result in spurious correlations and unstable predictive relationships, leading to poor performance of models when applied to out-of-sample (OOS) domains. To address these issues, we investigate \textit{Domain Generalization} techniques, with a particular focus on causal representation learning to improve a prediction model's generalizability to OOS domains. By leveraging multi-factor models from econometrics, we introduce a novel error bound that explicitly incorporates causal relationships. In addition, we present the connection between the proposed error bound and market nonstationarity. We also develop a \textit{Causal Discovery} technique to discover invariant feature representations, which effectively mitigates the proposed error bound, and the influence of spurious correlations on causal discovery is rigorously examined. Our theoretical findings are substantiated by numerical results, showcasing the effectiveness of our approach in enhancing the generalizability of stock prediction models.
Knowledge graphs have emerged to be promising datastore candidates for context augmentation during Retrieval Augmented Generation (RAG). As a result, techniques in graph representation learning have been simultaneously explored alongside principal neural information retrieval approaches, such as two-phased retrieval, also known as re-ranking. While Graph Neural Networks (GNNs) have been proposed to demonstrate proficiency in graph learning for re-ranking, there are ongoing limitations in modeling and evaluating input graph structures for training and evaluation for passage and document ranking tasks. In this survey, we review emerging GNN-based ranking model architectures along with their corresponding graph representation construction methodologies. We conclude by providing recommendations on future research based on community-wide challenges and opportunities.
In general, objects can be distinguished on the basis of their features, such as color or shape. In particular, it is assumed that similarity judgments about such features can be processed independently in different metric spaces. However, the unsupervised categorization mechanism of metric spaces corresponding to object features remains unknown. Here, we show that the artificial neural network system can autonomously categorize metric spaces through representation learning to satisfy the algebraic independence between neural networks, and project sensory information onto multiple high-dimensional metric spaces to independently evaluate the differences and similarities between features. Conventional methods often constrain the axes of the latent space to be mutually independent or orthogonal. However, the independent axes are not suitable for categorizing metric spaces. High-dimensional metric spaces that are independent of each other are not uniquely determined by the mutually independent axes, because any combination of independent axes can form mutually independent spaces. In other words, the mutually independent axes cannot be used to naturally categorize different feature spaces, such as color space and shape space. Therefore, constraining the axes to be mutually independent makes it difficult to categorize high-dimensional metric spaces. To overcome this problem, we developed a method to constrain only the spaces to be mutually independent and not the composed axes to be independent. Our theory provides general conditions for the unsupervised categorization of independent metric spaces, thus advancing the mathematical theory of functional differentiation of neural networks.
How can we accurately classify graphs? Graph classification is a pivotal task in data mining with applications in social network analysis, web analysis, drug discovery, molecular property prediction, etc. Graph neural networks have achieved the state-of-the-art performance in graph classification, but they consistently struggle with overfitting. To mitigate overfitting, researchers have introduced various representation learning methods utilizing graph augmentation. However, existing methods rely on simplistic use of graph augmentation, which loses augmentation-induced differences and limits the expressiveness of representations. In this paper, we propose AugWard (Augmentation-Aware Training with Graph Distance and Consistency Regularization), a novel graph representation learning framework that carefully considers the diversity introduced by graph augmentation. AugWard applies augmentation-aware training to predict the graph distance between the augmented graph and its original one, aligning the representation difference directly with graph distance at both feature and structure levels. Furthermore, AugWard employs consistency regularization to encourage the classifier to handle richer representations. Experimental results show that AugWard gives the state-of-the-art performance in supervised, semi-supervised graph classification, and transfer learning.
While sequential decision-making environments often involve high-dimensional observations, not all features of these observations are relevant for control. In particular, the observation space may capture factors of the environment which are not controllable by the agent, but which add complexity to the observation space. The need to ignore these "noise" features in order to operate in a tractably-small state space poses a challenge for efficient policy learning. Due to the abundance of video data available in many such environments, task-independent representation learning from action-free offline data offers an attractive solution. However, recent work has highlighted theoretical limitations in action-free learning under the Exogenous Block MDP (Ex-BMDP) model, where temporally-correlated noise features are present in the observations. To address these limitations, we identify a realistic setting where representation learning in Ex-BMDPs becomes tractable: when action-free video data from multiple agents with differing policies are available. Concretely, this paper introduces CRAFT (Comparison-based Representations from Action-Free Trajectories), a sample-efficient algorithm leveraging differences in controllable feature dynamics across agents to learn representations. We provide theoretical guarantees for CRAFT's performance and demonstrate its feasibility on a toy example, offering a foundation for practical methods in similar settings.
Human decision making can be challenging to predict because decisions are affected by a number of complex factors. Adding to this complexity, decision-making processes can differ considerably between individuals, and methods aimed at predicting human decisions need to take individual differences into account. Behavioral science offers methods by which to measure individual differences (e.g., questionnaires, behavioral models), but these are often narrowed down to low dimensions and not tailored to specific prediction tasks. This paper investigates the use of representation learning to measure individual differences from behavioral experiment data. Representation learning offers a flexible approach to create individual embeddings from data that are both structured (e.g., demographic information) and unstructured (e.g., free text), where the flexibility provides more options for individual difference measures for personalization, e.g., free text responses may allow for open-ended questions that are less privacy-sensitive. In the current paper we use representation learning to characterize individual differences in human performance on an economic decision-making task. We demonstrate that models using representation learning to capture individual differences consistently improve decision predictions over models without representation learning, and even outperform well-known theory-based behavioral models used in these environments. Our results propose that representation learning offers a useful and flexible tool to capture individual differences.
Understanding a procedural activity requires modeling both how action steps transform the scene, and how evolving scene transformations can influence the sequence of action steps, even those that are accidental or erroneous. Existing work has studied procedure-aware video representations by proposing novel approaches such as modeling the temporal order of actions and has not explicitly learned the state changes (scene transformations). In this work, we study procedure-aware video representation learning by incorporating state-change descriptions generated by Large Language Models (LLMs) as supervision signals for video encoders. Moreover, we generate state-change counterfactuals that simulate hypothesized failure outcomes, allowing models to learn by imagining the unseen ``What if'' scenarios. This counterfactual reasoning facilitates the model's ability to understand the cause and effect of each step in an activity. To verify the procedure awareness of our model, we conduct extensive experiments on procedure-aware tasks, including temporal action segmentation and error detection. Our results demonstrate the effectiveness of the proposed state-change descriptions and their counterfactuals and achieve significant improvements on multiple tasks. We will make our source code and data publicly available soon.