In recent advancements in unsupervised visual representation learning, the Joint-Embedding Predictive Architecture (JEPA) has emerged as a significant method for extracting visual features from unlabeled imagery through an innovative masking strategy. Despite its success, two primary limitations have been identified: the inefficacy of Exponential Moving Average (EMA) from I-JEPA in preventing entire collapse and the inadequacy of I-JEPA prediction in accurately learning the mean of patch representations.

Our contributions are manifold and significant. Firstly, we identify and articulate the limitations inherent in the I-JEP A framework, specifically its EMA and prediction mechanisms.

Self-Supervised Learning ( SSL) has revolutionized representation learning for speech and audio, enabling models to learn from unlabeled data and excel in diverse downstream tasks [ 1, 2, 3, 4 ] . Early SSL approaches for audio, such as contrastive predictive coding and wav2vec 2.0, learned latent speech representations by masking the input and solving a contrastive task over latent codes [ 5 ] . Follow-up methods like HuBERT [ 1 ] introduced offline clustering to generate pseudo-labels for masked audio segments and WavLM [ 6 ] applied data augmentation and denoising to improve robustness in speech representation learning. More recently, latent prediction approaches have gained traction: data2vec [ 7 ] and its efficient successor data2vec 2.0 [ 8 ] employ a teacher-student framework to predict contextualized latent representations of the input, achieving strong results across vision, speech, and language tasks. In the audio domain, Niizumi et al. introduced Masked Modeling Duo (M2D) [ 4 ], which uses two networks (online and momentum encoder) to predict masked patch embeddings and attained state-of-the-art results on numerous audio benchmarks. In computer vision, a new paradigm called Joint-Embedding Predictive Architecture (JEP A) [ 9, 10, 11 ] has been proposed to predict hidden content in a high-level latent space instead of pixel space.

In recent advancements in unsupervised visual representation learning, the Joint-Embedding Predictive Architecture (JEPA) has emerged as a significant method for extracting visual features from unlabeled imagery through an innovative masking strategy. Despite its success, two primary limitations have been identified: the inefficacy of Exponential Moving Average (EMA) from I-JEPA in preventing entire collapse and the inadequacy of I-JEPA prediction in accurately learning the mean of patch representations. This integration is designed to effectively learn the variance/covariance for preventing entire collapse and ensuring invariance in the mean of augmented views, thereby overcoming the identified limitations. Through empirical and theoretical evaluations, our work demonstrates that C-JEPA significantly enhances the stability and quality of visual representation learning. When pre-trained on the ImageNet-1K dataset, C-JEPA exhibits rapid and improved convergence in both linear probing and fine-tuning performance metrics.

Existing methods for self-supervised representation learning of geospatial regions and map entities rely extensively on the design of pretext tasks, often involving augmentations or heuristic sampling of positive and negative pairs based on spatial proximity. This reliance introduces biases and limits the representations' expressiveness and generalisability. Consequently, the literature has expressed a pressing need to explore different methods for modelling geospatial data. To address the key difficulties of such methods, namely multimodality, heterogeneity, and the choice of pretext tasks, we present GeoJEPA, a versatile multimodal fusion model for geospatial data built on the self-supervised Joint-Embedding Predictive Architecture. With GeoJEPA, we aim to eliminate the widely accepted augmentation- and sampling biases found in self-supervised geospatial representation learning. GeoJEPA uses self-supervised pretraining on a large dataset of OpenStreetMap attributes, geometries and aerial images. The results are multimodal semantic representations of urban regions and map entities that we evaluate both quantitatively and qualitatively. Through this work, we uncover several key insights into JEPA's ability to handle multimodal data.

In this paper, we tackle the task of musical stem retrieval. Given a musical mix, it consists in retrieving a stem that would fit with it, i.e., that would sound pleasant if played together. To do so, we introduce a new method based on Joint-Embedding Predictive Architectures, where an encoder and a predictor are jointly trained to produce latent representations of a context and predict latent representations of a target. In particular, we design our predictor to be conditioned on arbitrary instruments, enabling our model to perform zero-shot stem retrieval. In addition, we discover that pretraining the encoder using contrastive learning drastically improves the model's performance. We validate the retrieval performances of our model using the MUSDB18 and MoisesDB datasets. We show that it significantly outperforms previous baselines on both datasets, showcasing its ability to support more or less precise (and possibly unseen) conditioning. We also evaluate the learned embeddings on a beat tracking task, demonstrating that they retain temporal structure and local information.

In recent advancements in unsupervised visual representation learning, the Joint-Embedding Predictive Architecture (JEPA) has emerged as a significant method for extracting visual features from unlabeled imagery through an innovative masking strategy. Despite its success, two primary limitations have been identified: the inefficacy of Exponential Moving Average (EMA) from I-JEPA in preventing entire collapse and the inadequacy of I-JEPA prediction in accurately learning the mean of patch representations. Addressing these challenges, this study introduces a novel framework, namely C-JEPA (Contrastive-JEPA), which integrates the Image-based Joint-Embedding Predictive Architecture with the Variance-Invariance-Covariance Regularization (VICReg) strategy. This integration is designed to effectively learn the variance/covariance for preventing entire collapse and ensuring invariance in the mean of augmented views, thereby overcoming the identified limitations. Through empirical and theoretical evaluations, our work demonstrates that C-JEPA significantly enhances the stability and quality of visual representation learning. When pre-trained on the ImageNet-1K dataset, C-JEPA exhibits rapid and improved convergence in both linear probing and fine-tuning performance metrics.

In recent years, foundation models (FMs) such as large language models (LLMs) and latent diffusion models (LDMs) have profoundly impacted diverse sectors, including music. This comprehensive review examines state-of-the-art (SOTA) pre-trained models and foundation models in music, spanning from representation learning, generative learning and multimodal learning. We first contextualise the significance of music in various industries and trace the evolution of AI in music. By delineating the modalities targeted by foundation models, we discover many of the music representations are underexplored in FM development. Then, emphasis is placed on the lack of versatility of previous methods on diverse music applications, along with the potential of FMs in music understanding, generation and medical application. By comprehensively exploring the details of the model pre-training paradigm, architectural choices, tokenisation, finetuning methodologies and controllability, we emphasise the important topics that should have been well explored, like instruction tuning and in-context learning, scaling law and emergent ability, as well as long-sequence modelling etc. A dedicated section presents insights into music agents, accompanied by a thorough analysis of datasets and evaluations essential for pre-training and downstream tasks. Finally, by underscoring the vital importance of ethical considerations, we advocate that following research on FM for music should focus more on such issues as interpretability, transparency, human responsibility, and copyright issues. The paper offers insights into future challenges and trends on FMs for music, aiming to shape the trajectory of human-AI collaboration in the music realm.

This paper explores the automated process of determining stem compatibility by identifying audio recordings of single instruments that blend well with a given musical context. To tackle this challenge, we present Stem-JEPA, a novel Joint-Embedding Predictive Architecture (JEPA) trained on a multi-track dataset using a self-supervised learning approach. Our model comprises two networks: an encoder and a predictor, which are jointly trained to predict the embeddings of compatible stems from the embeddings of a given context, typically a mix of several instruments. Training a model in this manner allows its use in estimating stem compatibility - retrieving, aligning, or generating a stem to match a given mix - or for downstream tasks such as genre or key estimation, as the training paradigm requires the model to learn information related to timbre, harmony, and rhythm. We evaluate our model's performance on a retrieval task on the MUSDB18 dataset, testing its ability to find the missing stem from a mix and through a subjective user study. We also show that the learned embeddings capture temporal alignment information and, finally, evaluate the representations learned by our model on several downstream tasks, highlighting that they effectively capture meaningful musical features.

Trajectory similarity computation is an essential technique for analyzing moving patterns of spatial data across various applications such as traffic management, wildlife tracking, and location-based services. Modern methods often apply deep learning techniques to approximate heuristic metrics but struggle to learn more robust and generalized representations from the vast amounts of unlabeled trajectory data. Recent approaches focus on self-supervised learning methods such as contrastive learning, which have made significant advancements in trajectory representation learning. However, contrastive learning-based methods heavily depend on manually pre-defined data augmentation schemes, limiting the diversity of generated trajectories and resulting in learning from such variations in 2D Euclidean space, which prevents capturing high-level semantic variations. To address these limitations, we propose T-JEPA, a self-supervised trajectory similarity computation method employing Joint-Embedding Predictive Architecture (JEPA) to enhance trajectory representation learning. T-JEPA samples and predicts trajectory information in representation space, enabling the model to infer the missing components of trajectories at high-level semantics without relying on domain knowledge or manual effort. Extensive experiments conducted on three urban trajectory datasets and two Foursquare datasets demonstrate the effectiveness of T-JEPA in trajectory similarity computation.