Search ResultsShowing 1-20 of 228
-
arXiv:2505.06224 (Published 2025-05-09)
Towards a Unified Representation Evaluation Framework Beyond Downstream Tasks
Comments: Accepted at IJCNN 2025Categories: cs.LGDownstream probing has been the dominant method for evaluating model representations, an important process given the increasing prominence of self-supervised learning and foundation models. However, downstream probing primarily assesses the availability of task-relevant information in the model's latent space, overlooking attributes such as equivariance, invariance, and disentanglement, which contribute to the interpretability, adaptability, and utility of representations in real-world applications. While some attempts have been made to measure these qualities in representations, no unified evaluation framework with modular, generalizable, and interpretable metrics exists. In this paper, we argue for the importance of representation evaluation beyond downstream probing. We introduce a standardized protocol to quantify informativeness, equivariance, invariance, and disentanglement of factors of variation in model representations. We use it to evaluate representations from a variety of models in the image and speech domains using different architectures and pretraining approaches on identified controllable factors of variation. We find that representations from models with similar downstream performance can behave substantially differently with regard to these attributes. This hints that the respective mechanisms underlying their downstream performance are functionally different, prompting new research directions to understand and improve representations.
-
arXiv:2505.03703 (Published 2025-05-06)
Fill the Gap: Quantifying and Reducing the Modality Gap in Image-Text Representation Learning
Vision-language models (VLMs) allow to embed texts and images in a shared representation space. However, it has been shown that these models are subject to a modality gap phenomenon meaning there exists a clear separation between the embeddings from one modality and another in the embedding space. While this misalignment is detrimental for downstream tasks such as multimodal retrieval, multimodal clustering or zero-shot classification, etc. no generic and practical methods have so far been proposed to assess it precisely and even reduce it. We therefore propose novel measures and effective techniques (spectral- and optimal transport-based methods) to achieve this goal. Extensive experiments conducted on several image-text datasets and models demonstrate their effectiveness and beneficial effects on downstream tasks. Our code is available at the URL provided in the paper's abstract.
-
arXiv:2502.08918 (Published 2025-02-13)
CLEAR: Cluster-based Prompt Learning on Heterogeneous Graphs
Comments: accepted by PAKDD 2025Categories: cs.LGPrompt learning has attracted increasing attention in the graph domain as a means to bridge the gap between pretext and downstream tasks. Existing studies on heterogeneous graph prompting typically use feature prompts to modify node features for specific downstream tasks, which do not concern the structure of heterogeneous graphs. Such a design also overlooks information from the meta-paths, which are core to learning the high-order semantics of the heterogeneous graphs. To address these issues, we propose CLEAR, a Cluster-based prompt LEARNING model on heterogeneous graphs. We present cluster prompts that reformulate downstream tasks as heterogeneous graph reconstruction. In this way, we align the pretext and downstream tasks to share the same training objective. Additionally, our cluster prompts are also injected into the meta-paths such that the prompt learning process incorporates high-order semantic information entailed by the meta-paths. Extensive experiments on downstream tasks confirm the superiority of CLEAR. It consistently outperforms state-of-the-art models, achieving up to 5% improvement on the F1 metric for node classification.
-
Exploratory Diffusion Model for Unsupervised Reinforcement Learning
Unsupervised reinforcement learning (URL) aims to pre-train agents by exploring diverse states or skills in reward-free environments, facilitating efficient adaptation to downstream tasks. As the agent cannot access extrinsic rewards during unsupervised exploration, existing methods design intrinsic rewards to model the explored data and encourage further exploration. However, the explored data are always heterogeneous, posing the requirements of powerful representation abilities for both intrinsic reward models and pre-trained policies. In this work, we propose the Exploratory Diffusion Model (ExDM), which leverages the strong expressive ability of diffusion models to fit the explored data, simultaneously boosting exploration and providing an efficient initialization for downstream tasks. Specifically, ExDM can accurately estimate the distribution of collected data in the replay buffer with the diffusion model and introduces the score-based intrinsic reward, encouraging the agent to explore less-visited states. After obtaining the pre-trained policies, ExDM enables rapid adaptation to downstream tasks. In detail, we provide theoretical analyses and practical algorithms for fine-tuning diffusion policies, addressing key challenges such as training instability and computational complexity caused by multi-step sampling. Extensive experiments demonstrate that ExDM outperforms existing SOTA baselines in efficient unsupervised exploration and fast fine-tuning downstream tasks, especially in structurally complicated environments.
-
arXiv:2502.04465 (Published 2025-02-06)
FocalCodec: Low-Bitrate Speech Coding via Focal Modulation Networks
Comments: 18 pagesLarge language models have revolutionized natural language processing through self-supervised pretraining on massive datasets. Inspired by this success, researchers have explored adapting these methods to speech by discretizing continuous audio into tokens using neural audio codecs. However, existing approaches face limitations, including high bitrates, the loss of either semantic or acoustic information, and the reliance on multi-codebook designs when trying to capture both, which increases architectural complexity for downstream tasks. To address these challenges, we introduce FocalCodec, an efficient low-bitrate codec based on focal modulation that utilizes a single binary codebook to compress speech between 0.16 and 0.65 kbps. FocalCodec delivers competitive performance in speech resynthesis and voice conversion at lower bitrates than the current state-of-the-art, while effectively handling multilingual speech and noisy environments. Evaluation on downstream tasks shows that FocalCodec successfully preserves sufficient semantic and acoustic information, while also being well-suited for generative modeling. Demo samples, code and checkpoints are available at https://lucadellalib.github.io/focalcodec-web/.
-
arXiv:2502.03439 (Published 2025-02-05)
Linearized Optimal Transport pyLOT Library: A Toolkit for Machine Learning on Point Clouds
Keywords: linearized optimal transport pylot library, machine learning, point clouds, pipeline embeds probability distributions, downstream tasksTags: research toolThe pyLOT library offers a Python implementation of linearized optimal transport (LOT) techniques and methods to use in downstream tasks. The pipeline embeds probability distributions into a Hilbert space via the Optimal Transport maps from a fixed reference distribution, and this linearization allows downstream tasks to be completed using off the shelf (linear) machine learning algorithms. We provide a case study of performing ML on 3D scans of lemur teeth, where the original questions of classification, clustering, dimension reduction, and data generation reduce to simple linear operations performed on the LOT embedded representations.
-
arXiv:2501.18824 (Published 2025-01-31)
Memory-Efficient Fine-Tuning of Transformers via Token Selection
Comments: EMNLP 2024Fine-tuning provides an effective means to specialize pre-trained models for various downstream tasks. However, fine-tuning often incurs high memory overhead, especially for large transformer-based models, such as LLMs. While existing methods may reduce certain parts of the memory required for fine-tuning, they still require caching all intermediate activations computed in the forward pass to update weights during the backward pass. In this work, we develop TokenTune, a method to reduce memory usage, specifically the memory to store intermediate activations, in the fine-tuning of transformer-based models. During the backward pass, TokenTune approximates the gradient computation by backpropagating through just a subset of input tokens. Thus, with TokenTune, only a subset of intermediate activations are cached during the forward pass. Also, TokenTune can be easily combined with existing methods like LoRA, further reducing the memory cost. We evaluate our approach on pre-trained transformer models with up to billions of parameters, considering the performance on multiple downstream tasks such as text classification and question answering in a few-shot learning setup. Overall, TokenTune achieves performance on par with full fine-tuning or representative memory-efficient fine-tuning methods, while greatly reducing the memory footprint, especially when combined with other methods with complementary memory reduction mechanisms. We hope that our approach will facilitate the fine-tuning of large transformers, in specializing them for specific domains or co-training them with other neural components from a larger system. Our code is available at https://github.com/facebookresearch/tokentune.
-
arXiv:2501.18739 (Published 2025-01-30)
Neural Graph Pattern Machine
Graph learning tasks require models to comprehend essential substructure patterns relevant to downstream tasks, such as triadic closures in social networks and benzene rings in molecular graphs. Due to the non-Euclidean nature of graphs, existing graph neural networks (GNNs) rely on message passing to iteratively aggregate information from local neighborhoods. Despite their empirical success, message passing struggles to identify fundamental substructures, such as triangles, limiting its expressiveness. To overcome this limitation, we propose the Neural Graph Pattern Machine (GPM), a framework designed to learn directly from graph patterns. GPM efficiently extracts and encodes substructures while identifying the most relevant ones for downstream tasks. We also demonstrate that GPM offers superior expressivity and improved long-range information modeling compared to message passing. Empirical evaluations on node classification, link prediction, graph classification, and regression show the superiority of GPM over state-of-the-art baselines. Further analysis reveals its desirable out-of-distribution robustness, scalability, and interpretability. We consider GPM to be a step toward going beyond message passing.
-
arXiv:2501.15142 (Published 2025-01-25)
DAGPrompT: Pushing the Limits of Graph Prompting with a Distribution-aware Graph Prompt Tuning Approach
Comments: To be published in WWW '25, April 28-May 2, 2025, Sydney, NSW, AustraliaKeywords: distribution-aware graph prompt tuning approach, graph prompting methods, downstream tasks, gnn encoders projection matrix, complex graphsTags: journal articleThe pre-train then fine-tune approach has advanced GNNs by enabling general knowledge capture without task-specific labels. However, an objective gap between pre-training and downstream tasks limits its effectiveness. Recent graph prompting methods aim to close this gap through task reformulations and learnable prompts. Despite this, they struggle with complex graphs like heterophily graphs. Freezing the GNN encoder can reduce the impact of prompting, while simple prompts fail to handle diverse hop-level distributions. This paper identifies two key challenges in adapting graph prompting methods for complex graphs: (1) adapting the model to new distributions in downstream tasks to mitigate pre-training and fine-tuning discrepancies from heterophily and (2) customizing prompts for hop-specific node requirements. To overcome these challenges, we propose Distribution-aware Graph Prompt Tuning (DAGPrompT), which integrates a GLoRA module for optimizing the GNN encoder's projection matrix and message-passing schema through low-rank adaptation. DAGPrompT also incorporates hop-specific prompts accounting for varying graph structures and distributions among hops. Evaluations on 10 datasets and 14 baselines demonstrate that DAGPrompT improves accuracy by up to 4.79 in node and graph classification tasks, setting a new state-of-the-art while preserving efficiency. Codes are available at GitHub.
-
arXiv:2501.14685 (Published 2025-01-24)
Rethinking Foundation Models for Medical Image Classification through a Benchmark Study on MedMNIST
Comments: submitted to MIDL2025Foundation models are widely employed in medical image analysis, due to their high adaptability and generalizability for downstream tasks. With the increasing number of foundation models being released, model selection has become an important issue. In this work, we study the capabilities of foundation models in medical image classification tasks by conducting a benchmark study on the MedMNIST dataset. Specifically, we adopt various foundation models ranging from convolutional to Transformer-based models and implement both end-to-end training and linear probing for all classification tasks. The results demonstrate the significant potential of these pre-trained models when transferred for medical image classification. We further conduct experiments with different image sizes and various sizes of training data. By analyzing all the results, we provide preliminary, yet useful insights and conclusions on this topic.
-
Handling Missing Data in Downstream Tasks With Distribution-Preserving Guarantees
Categories: cs.LGMissing feature values are a significant hurdle for downstream machine-learning tasks such as classification. However, imputation methods for classification might be time-consuming for high-dimensional data, and offer few theoretical guarantees on the preservation of the data distribution and imputation quality, especially for not-missing-at-random mechanisms. First, we propose an imputation approach named F3I based on the iterative improvement of a K-nearest neighbor imputation, where neighbor-specific weights are learned through the optimization of a novel concave, differentiable objective function related to the preservation of the data distribution on non-missing values. F3I can then be chained to and jointly trained with any classifier architecture. Second, we provide a theoretical analysis of imputation quality and data distribution preservation by F3I for several types of missing mechanisms. Finally, we demonstrate the superior performance of F3I on several imputation and classification tasks, with applications to drug repurposing and handwritten-digit recognition data.
-
arXiv:2501.04641 (Published 2025-01-08)
A Statistical Theory of Contrastive Pre-training and Multimodal Generative AI
Comments: 108 pagesMulti-modal generative AI systems, such as those combining vision and language, rely on contrastive pre-training to learn representations across different modalities. While their practical benefits are widely acknowledged, a rigorous theoretical understanding of the contrastive pre-training framework remains limited. This paper develops a theoretical framework to explain the success of contrastive pre-training in downstream tasks, such as zero-shot classification, conditional diffusion models, and vision-language models. We introduce the concept of approximate sufficient statistics, a generalization of the classical sufficient statistics, and show that near-minimizers of the contrastive pre-training loss are approximately sufficient, making them adaptable to diverse downstream tasks. We further propose the Joint Generative Hierarchical Model for the joint distribution of images and text, showing that transformers can efficiently approximate relevant functions within this model via belief propagation. Building on this framework, we derive sample complexity guarantees for multi-modal learning based on contrastive pre-trained representations. Numerical simulations validate these theoretical findings, demonstrating the strong generalization performance of contrastively pre-trained transformers in various multi-modal tasks.
-
arXiv:2501.02728 (Published 2025-01-06)
OpenGU: A Comprehensive Benchmark for Graph Unlearning
Comments: under reviewGraph Machine Learning is essential for understanding and analyzing relational data. However, privacy-sensitive applications demand the ability to efficiently remove sensitive information from trained graph neural networks (GNNs), avoiding the unnecessary time and space overhead caused by retraining models from scratch. To address this issue, Graph Unlearning (GU) has emerged as a critical solution, with the potential to support dynamic graph updates in data management systems and enable scalable unlearning in distributed data systems while ensuring privacy compliance. Unlike machine unlearning in computer vision or other fields, GU faces unique difficulties due to the non-Euclidean nature of graph data and the recursive message-passing mechanism of GNNs. Additionally, the diversity of downstream tasks and the complexity of unlearning requests further amplify these challenges. Despite the proliferation of diverse GU strategies, the absence of a benchmark providing fair comparisons for GU, and the limited flexibility in combining downstream tasks and unlearning requests, have yielded inconsistencies in evaluations, hindering the development of this domain. To fill this gap, we present OpenGU, the first GU benchmark, where 16 SOTA GU algorithms and 37 multi-domain datasets are integrated, enabling various downstream tasks with 13 GNN backbones when responding to flexible unlearning requests. Based on this unified benchmark framework, we are able to provide a comprehensive and fair evaluation for GU. Through extensive experimentation, we have drawn $8$ crucial conclusions about existing GU methods, while also gaining valuable insights into their limitations, shedding light on potential avenues for future research.
-
arXiv:2501.00709 (Published 2025-01-01)
KAN KAN Buff Signed Graph Neural Networks?
Categories: cs.LGGraph Representation Learning aims to create embeddings for nodes and edges, capturing their features and interconnections. Graph Neural Networks (GNNs) have excelled in this task, leveraging neural networks to model complex graph relationships. Recently, the Kolmogorov-Arnold Neural Network (KAN) emerged as an alternative to Multi-Layer Perceptron (MLP), showing improved accuracy and interpretability with fewer parameters. While KANs have been integrated into unsigned GNNs, their application in signed GNNs remains unexplored. This paper integrates KAN into Signed Graph Convolutional Networks (SGCNs) to evaluate its performance on signed graphs where edges have positive or negative signs. We empirically assess KAN-enhanced SGCNs (KASGCN) on downstream tasks such as signed community detection and link sign prediction to enhance the embedding quality in signed networks. Considering the variability in the results indicated by the relatively large standard deviation, KASGCN demonstrates competitive performance with, or similar to, the vanilla SGCN in the evaluated downstream tasks, and its effectiveness is context-dependent (signed graph and parameters...etc.).
-
arXiv:2412.03471 (Published 2024-12-04)
Cluster Specific Representation Learning
Categories: cs.LGRepresentation learning aims to extract meaningful lower-dimensional embeddings from data, known as representations. Despite its widespread application, there is no established definition of a ``good'' representation. Typically, the representation quality is evaluated based on its performance in downstream tasks such as clustering, de-noising, etc. However, this task-specific approach has a limitation where a representation that performs well for one task may not necessarily be effective for another. This highlights the need for a more agnostic formulation, which is the focus of our work. We propose a downstream-agnostic formulation: when inherent clusters exist in the data, the representations should be specific to each cluster. Under this idea, we develop a meta-algorithm that jointly learns cluster-specific representations and cluster assignments. As our approach is easy to integrate with any representation learning framework, we demonstrate its effectiveness in various setups, including Autoencoders, Variational Autoencoders, Contrastive learning models, and Restricted Boltzmann Machines. We qualitatively compare our cluster-specific embeddings to standard embeddings and downstream tasks such as de-noising and clustering. While our method slightly increases runtime and parameters compared to the standard model, the experiments clearly show that it extracts the inherent cluster structures in the data, resulting in improved performance in relevant applications.
-
arXiv:2411.15096 (Published 2024-11-22)
RED: Effective Trajectory Representation Learning with Comprehensive Information
Comments: This paper is accepted by VLDB2025Trajectory representation learning (TRL) maps trajectories to vectors that can then be used for various downstream tasks, including trajectory similarity computation, trajectory classification, and travel-time estimation. However, existing TRL methods often produce vectors that, when used in downstream tasks, yield insufficiently accurate results. A key reason is that they fail to utilize the comprehensive information encompassed by trajectories. We propose a self-supervised TRL framework, called RED, which effectively exploits multiple types of trajectory information. Overall, RED adopts the Transformer as the backbone model and masks the constituting paths in trajectories to train a masked autoencoder (MAE). In particular, RED considers the moving patterns of trajectories by employing a Road-aware masking strategy} that retains key paths of trajectories during masking, thereby preserving crucial information of the trajectories. RED also adopts a spatial-temporal-user joint Embedding scheme to encode comprehensive information when preparing the trajectories as model inputs. To conduct training, RED adopts Dual-objective task learning}: the Transformer encoder predicts the next segment in a trajectory, and the Transformer decoder reconstructs the entire trajectory. RED also considers the spatial-temporal correlations of trajectories by modifying the attention mechanism of the Transformer. We compare RED with 9 state-of-the-art TRL methods for 4 downstream tasks on 3 real-world datasets, finding that RED can usually improve the accuracy of the best-performing baseline by over 5%.
-
arXiv:2411.12074 (Published 2024-11-18)
Mitigating Gender Bias in Contextual Word Embeddings
Word embeddings have been shown to produce remarkable results in tackling a vast majority of NLP related tasks. Unfortunately, word embeddings also capture the stereotypical biases that are prevalent in society, affecting the predictive performance of the embeddings when used in downstream tasks. While various techniques have been proposed \cite{bolukbasi2016man, zhao2018learning} and criticized\cite{gonen2019lipstick} for static embeddings, very little work has focused on mitigating bias in contextual embeddings. In this paper, we propose a novel objective function for MLM(Masked-Language Modeling) which largely mitigates the gender bias in contextual embeddings and also preserves the performance for downstream tasks. Since previous works on measuring bias in contextual embeddings lack in normative reasoning, we also propose novel evaluation metrics that are straight-forward and aligned with our motivations in debiasing. We also propose new methods for debiasing static embeddings and provide empirical proof via extensive analysis and experiments, as to why the main source of bias in static embeddings stems from the presence of stereotypical names rather than gendered words themselves. All experiments and embeddings studied are in English, unless otherwise specified.\citep{bender2011achieving}.
-
arXiv:2411.09849 (Published 2024-11-14)
Self-Supervised Radio Pre-training: Toward Foundational Models for Spectrogram Learning
Foundational deep learning (DL) models are general models, trained on large, diverse, and unlabelled datasets, typically using self-supervised learning techniques have led to significant advancements especially in natural language processing. These pretrained models can be fine-tuned for related downstream tasks, offering faster development and reduced training costs, while often achieving improved performance. In this work, we introduce Masked Spectrogram Modeling, a novel self-supervised learning approach for pretraining foundational DL models on radio signals. Adopting a Convolutional LSTM architecture for efficient spatio-temporal processing, we pretrain the model with an unlabelled radio dataset collected from over-the-air measurements. Subsequently, the pretrained model is fine-tuned for two downstream tasks: spectrum forecasting and segmentation. Experimental results demonstrate that our methodology achieves competitive performance in both forecasting accuracy and segmentation, validating its effectiveness for developing foundational radio models.
-
arXiv:2411.07207 (Published 2024-11-11)
General Geospatial Inference with a Population Dynamics Foundation Model
Mohit Agarwal et al.Comments: 28 pages, 16 figures, preprintSupporting the health and well-being of dynamic populations around the world requires governmental agencies, organizations and researchers to understand and reason over complex relationships between human behavior and local contexts in order to identify high-risk groups and strategically allocate limited resources. Traditional approaches to these classes of problems often entail developing manually curated, task-specific features and models to represent human behavior and the natural and built environment, which can be challenging to adapt to new, or even, related tasks. To address this, we introduce a Population Dynamics Foundation Model (PDFM) that aims to capture the relationships between diverse data modalities and is applicable to a broad range of geospatial tasks. We first construct a geo-indexed dataset for postal codes and counties across the United States, capturing rich aggregated information on human behavior from maps, busyness, and aggregated search trends, and environmental factors such as weather and air quality. We then model this data and the complex relationships between locations using a graph neural network, producing embeddings that can be adapted to a wide range of downstream tasks using relatively simple models. We evaluate the effectiveness of our approach by benchmarking it on 27 downstream tasks spanning three distinct domains: health indicators, socioeconomic factors, and environmental measurements. The approach achieves state-of-the-art performance on all 27 geospatial interpolation tasks, and on 25 out of the 27 extrapolation and super-resolution tasks. We combined the PDFM with a state-of-the-art forecasting foundation model, TimesFM, to predict unemployment and poverty, achieving performance that surpasses fully supervised forecasting. The full set of embeddings and sample code are publicly available for researchers.
-
arXiv:2411.01800 (Published 2024-11-04)
Expanding Sparse Tuning for Low Memory Usage
Comments: Accepted by NeurIPS 2024Parameter-efficient fine-tuning (PEFT) is an effective method for adapting pre-trained vision models to downstream tasks by tuning a small subset of parameters. Among PEFT methods, sparse tuning achieves superior performance by only adjusting the weights most relevant to downstream tasks, rather than densely tuning the whole weight matrix. However, this performance improvement has been accompanied by increases in memory usage, which stems from two factors, i.e., the storage of the whole weight matrix as learnable parameters in the optimizer and the additional storage of tunable weight indexes. In this paper, we propose a method named SNELL (Sparse tuning with kerNELized LoRA) for sparse tuning with low memory usage. To achieve low memory usage, SNELL decomposes the tunable matrix for sparsification into two learnable low-rank matrices, saving from the costly storage of the whole original matrix. A competition-based sparsification mechanism is further proposed to avoid the storage of tunable weight indexes. To maintain the effectiveness of sparse tuning with low-rank matrices, we extend the low-rank decomposition by applying nonlinear kernel functions to the whole-matrix merging. Consequently, we gain an increase in the rank of the merged matrix, enhancing the ability of SNELL in adapting the pre-trained models to downstream tasks. Extensive experiments on multiple downstream tasks show that SNELL achieves state-of-the-art performance with low memory usage, endowing PEFT with sparse tuning to large-scale models. Codes are available at https://github.com/ssfgunner/SNELL.