Recent work has proven that training large language models with self-supervised tasks and fine-tuning these models to complete new tasks in a transfer learning setting is a powerful idea, enabling the creation of models with many parameters, even with little labeled data; however, the number of domains that have harnessed these advancements has been limited. In this work, we formulate a set of geometric tasks suitable for the large-scale study of ordered three-dimensional structures, without requiring any human intervention in data labeling. We build deep rotation- and permutation-equivariant neural networks based on geometric algebra and use them to solve these tasks on both idealized and simulated three-dimensional structures. Quantifying order in complex-structured assemblies remains a long-standing challenge in materials physics; these models can elucidate the behavior of real self-assembling systems in a variety of ways, from distilling insights from learned tasks without further modification to solving new tasks with smaller amounts of labeled data via transfer learning.
Test-time adaptation (TTA) of 3D point clouds is crucial for mitigating discrepancies between training and testing samples in real-world scenarios, particularly when handling corrupted point clouds. LiDAR data, for instance, can be affected by sensor failures or environmental factors, causing domain gaps. Adapting models to these distribution shifts online is crucial, as training for every possible variation is impractical. Existing methods often focus on fine-tuning pre-trained models based on self-supervised learning or pseudo-labeling, which can lead to forgetting valuable source domain knowledge over time and reduce generalization on future tests. In this paper, we introduce a novel 3D test-time adaptation method, termed 3DD-TTA, which stands for 3D Denoising Diffusion Test-Time Adaptation. This method uses a diffusion strategy that adapts input point cloud samples to the source domain while keeping the source model parameters intact. The approach uses a Variational Autoencoder (VAE) to encode the corrupted point cloud into a shape latent and latent points. These latent points are corrupted with Gaussian noise and subjected to a denoising diffusion process. During this process, both the shape latent and latent points are updated to preserve fidelity, guiding the denoising toward generating consistent samples that align more closely with the source domain. We conduct extensive experiments on the ShapeNet dataset and investigate its generalizability on ModelNet40 and ScanObjectNN, achieving state-of-the-art results. The code has been released at \url{https://github.com/hamidreza-dastmalchi/3DD-TTA}.
Spoken term detection (STD) is often hindered by reliance on frame-level features and the computationally intensive DTW-based template matching, limiting its practicality. To address these challenges, we propose a novel approach that encodes speech into discrete, speaker-agnostic semantic tokens. This facilitates fast retrieval using text-based search algorithms and effectively handles out-of-vocabulary terms. Our approach focuses on generating consistent token sequences across varying utterances of the same term. We also propose a bidirectional state space modeling within the Mamba encoder, trained in a self-supervised learning framework, to learn contextual frame-level features that are further encoded into discrete tokens. Our analysis shows that our speech tokens exhibit greater speaker invariance than those from existing tokenizers, making them more suitable for STD tasks. Empirical evaluation on LibriSpeech and TIMIT databases indicates that our method outperforms existing STD baselines while being more efficient.
Molecular featurisation refers to the transformation of molecular data into numerical feature vectors. It is one of the key research areas in molecular machine learning and computational drug discovery. Recently, message-passing graph neural networks (GNNs) have emerged as a novel method to learn differentiable features directly from molecular graphs. While such techniques hold great promise, further investigations are needed to clarify if and when they indeed manage to definitively outcompete classical molecular featurisations such as extended-connectivity fingerprints (ECFPs) and physicochemical-descriptor vectors (PDVs). We systematically explore and further develop classical and graph-based molecular featurisation methods for two important tasks: molecular property prediction, in particular, quantitative structure-activity relationship (QSAR) prediction, and the largely unexplored challenge of activity-cliff (AC) prediction. We first give a technical description and critical analysis of PDVs, ECFPs and message-passing GNNs, with a focus on graph isomorphism networks (GINs). We then conduct a rigorous computational study to compare the performance of PDVs, ECFPs and GINs for QSAR and AC-prediction. Following this, we mathematically describe and computationally evaluate a novel twin neural network model for AC-prediction. We further introduce an operation called substructure pooling for the vectorisation of structural fingerprints as a natural counterpart to graph pooling in GNN architectures. We go on to propose Sort & Slice, a simple substructure-pooling technique for ECFPs that robustly outperforms hash-based folding at molecular property prediction. Finally, we outline two ideas for future research: (i) a graph-based self-supervised learning strategy to make classical molecular featurisations trainable, and (ii) trainable substructure-pooling via differentiable self-attention.
Representation learning of pathology whole-slide images(WSIs) has primarily relied on weak supervision with Multiple Instance Learning (MIL). This approach leads to slide representations highly tailored to a specific clinical task. Self-supervised learning (SSL) has been successfully applied to train histopathology foundation models (FMs) for patch embedding generation. However, generating patient or slide level embeddings remains challenging. Existing approaches for slide representation learning extend the principles of SSL from patch level learning to entire slides by aligning different augmentations of the slide or by utilizing multimodal data. By integrating tile embeddings from multiple FMs, we propose a new single modality SSL method in feature space that generates useful slide representations. Our contrastive pretraining strategy, called COBRA, employs multiple FMs and an architecture based on Mamba-2. COBRA exceeds performance of state-of-the-art slide encoders on four different public Clinical Protemic Tumor Analysis Consortium (CPTAC) cohorts on average by at least +4.5% AUC, despite only being pretrained on 3048 WSIs from The Cancer Genome Atlas (TCGA). Additionally, COBRA is readily compatible at inference time with previously unseen feature extractors. Code available at https://github.com/KatherLab/COBRA.
The upcoming Square Kilometer Array (SKA) telescope marks a significant step forward in radio astronomy, presenting new opportunities and challenges for data analysis. Traditional visual models pretrained on optical photography images may not perform optimally on radio interferometry images, which have distinct visual characteristics. Self-Supervised Learning (SSL) offers a promising approach to address this issue, leveraging the abundant unlabeled data in radio astronomy to train neural networks that learn useful representations from radio images. This study explores the application of SSL to radio astronomy, comparing the performance of SSL-trained models with that of traditional models pretrained on natural images, evaluating the importance of data curation for SSL, and assessing the potential benefits of self-supervision to different domain-specific radio astronomy datasets. Our results indicate that, SSL-trained models achieve significant improvements over the baseline in several downstream tasks, especially in the linear evaluation setting; when the entire backbone is fine-tuned, the benefits of SSL are less evident but still outperform pretraining. These findings suggest that SSL can play a valuable role in efficiently enhancing the analysis of radio astronomical data. The trained models and code is available at: \url{https://github.com/dr4thmos/solo-learn-radio}
Group Point-of-Interest (POI) recommendations aim to predict the next POI that satisfies the diverse preferences of a group of users. This task is more challenging than traditional individual POI recommendations due to complex group decision-making and extremely sparse group-level check-in data. Existing methods for group POI recommendations primarily rely on single ID-based features from check-in data, capturing only statistical correlations and failing to fully utilize the rich semantic information contained in the check-ins, resulting in suboptimal performance. To this end, we propose a framework that unleashes the power of the Large Language Model (LLM) for context-aware group POI recommendations (LLMGPR). Our approach first introduces POI tokens alongside the original word tokens of the LLM, which are initialized by applying the LLM to the rich information of each POI. We then propose a novel sequencing adapter guided by Quantized Low-Rank Adaptation (QLORA) to modify the LLM. The enhanced LLM can learn sequence representations by combining semantic-enhanced POI tokens and rich contextual information including positional encodings and spatio-temporal differences. This approach can be adapted for learning either group or user representations depending on the sequence type. Furthermore, we enhance group representations by aggregating individual member representations with another QLORA-based aggregation adapter and introducing a self-supervised learning task that predicts the purpose of check-in sequences, alleviating the data sparsity issue. Our experimental results demonstrate that LLMGPR outperforms existing methods, effectively addressing group-level data sparsity and providing superior recommendations.
Deep learning has proven very promising for interpreting MRI in brain tumor diagnosis. However, deep learning models suffer from a scarcity of brain MRI datasets for effective training. Self-supervised learning (SSL) models provide data-efficient and remarkable solutions to limited dataset problems. Therefore, this paper introduces a generative SSL model for brain tumor classification in two stages. The first stage is designed to pre-train a Residual Vision Transformer (ResViT) model for MRI synthesis as a pretext task. The second stage includes fine-tuning a ResViT-based classifier model as a downstream task. Accordingly, we aim to leverage local features via CNN and global features via ViT, employing a hybrid CNN-transformer architecture for ResViT in pretext and downstream tasks. Moreover, synthetic MRI images are utilized to balance the training set. The proposed model performs on public BraTs 2023, Figshare, and Kaggle datasets. Furthermore, we compare the proposed model with various deep learning models, including A-UNet, ResNet-9, pix2pix, pGAN for MRI synthesis, and ConvNeXtTiny, ResNet101, DenseNet12, Residual CNN, ViT for classification. According to the results, the proposed model pretraining on the MRI dataset is superior compared to the pretraining on the ImageNet dataset. Overall, the proposed model attains the highest accuracy, achieving 90.56% on the BraTs dataset with T1 sequence, 98.53% on the Figshare, and 98.47% on the Kaggle brain tumor datasets. As a result, the proposed model demonstrates a robust, effective, and successful approach to handling insufficient dataset challenges in MRI analysis by incorporating SSL, fine-tuning, data augmentation, and combining CNN and ViT.
Self-supervised learning has made substantial strides in image processing, while visual pre-training for autonomous driving is still in its infancy. Existing methods often focus on learning geometric scene information while neglecting texture or treating both aspects separately, hindering comprehensive scene understanding. In this context, we are excited to introduce GaussianPretrain, a novel pre-training paradigm that achieves a holistic understanding of the scene by uniformly integrating geometric and texture representations. Conceptualizing 3D Gaussian anchors as volumetric LiDAR points, our method learns a deepened understanding of scenes to enhance pre-training performance with detailed spatial structure and texture, achieving that 40.6% faster than NeRF-based method UniPAD with 70% GPU memory only. We demonstrate the effectiveness of GaussianPretrain across multiple 3D perception tasks, showing significant performance improvements, such as a 7.05% increase in NDS for 3D object detection, boosts mAP by 1.9% in HD map construction and 0.8% improvement on Occupancy prediction. These significant gains highlight GaussianPretrain's theoretical innovation and strong practical potential, promoting visual pre-training development for autonomous driving. Source code will be available at https://github.com/Public-BOTs/GaussianPretrain
The disperse structure distributions (discreteness) and variant scattering characteristics (variability) of SAR airplane targets lead to special challenges of object detection and recognition. The current deep learning-based detectors encounter challenges in distinguishing fine-grained SAR airplanes against complex backgrounds. To address it, we propose a novel physics-guided detector (PGD) learning paradigm for SAR airplanes that comprehensively investigate their discreteness and variability to improve the detection performance. It is a general learning paradigm that can be extended to different existing deep learning-based detectors with "backbone-neck-head" architectures. The main contributions of PGD include the physics-guided self-supervised learning, feature enhancement, and instance perception, denoted as PGSSL, PGFE, and PGIP, respectively. PGSSL aims to construct a self-supervised learning task based on a wide range of SAR airplane targets that encodes the prior knowledge of various discrete structure distributions into the embedded space. Then, PGFE enhances the multi-scale feature representation of a detector, guided by the physics-aware information learned from PGSSL. PGIP is constructed at the detection head to learn the refined and dominant scattering point of each SAR airplane instance, thus alleviating the interference from the complex background. We propose two implementations, denoted as PGD and PGD-Lite, and apply them to various existing detectors with different backbones and detection heads. The experiments demonstrate the flexibility and effectiveness of the proposed PGD, which can improve existing detectors on SAR airplane detection with fine-grained classification task (an improvement of 3.1\% mAP most), and achieve the state-of-the-art performance (90.7\% mAP) on SAR-AIRcraft-1.0 dataset. The project is open-source at \url{https://github.com/XAI4SAR/PGD}.
In this work, we attempted to extend the thought and showcase a way forward for the Self-supervised Learning (SSL) learning paradigm by combining contrastive learning, self-distillation (knowledge distillation) and masked data modelling, the three major SSL frameworks, to learn a joint and coordinated representation. The proposed technique of SSL learns by the collaborative power of different learning objectives of SSL. Hence to jointly learn the different SSL objectives we proposed a new SSL architecture KDC-MAE, a complementary masking strategy to learn the modular correspondence, and a weighted way to combine them coordinately. Experimental results conclude that the contrastive masking correspondence along with the KD learning objective has lent a hand to performing better learning for multiple modalities over multiple tasks.
This study aims to optimize the few-shot image classification task and improve the model's feature extraction and classification performance by combining self-supervised learning with the deep network model ResNet-101. During the training process, we first pre-train the model with self-supervision to enable it to learn common feature expressions on a large amount of unlabeled data; then fine-tune it on the few-shot dataset Mini-ImageNet to improve the model's accuracy and generalization ability under limited data. The experimental results show that compared with traditional convolutional neural networks, ResNet-50, DenseNet, and other models, our method has achieved excellent performance of about 95.12% in classification accuracy (ACC) and F1 score, verifying the effectiveness of self-supervised learning in few-shot classification. This method provides an efficient and reliable solution for the field of few-shot image classification.
Directed Energy Deposition (DED) offers significant potential for manufacturing complex and multi-material parts. However, internal defects such as porosity and cracks can compromise mechanical properties and overall performance. This study focuses on in-situ monitoring and characterization of melt pools associated with porosity, aiming to improve defect detection and quality control in DED-printed parts. Traditional machine learning approaches for defect identification rely on extensive labeled datasets, often scarce and expensive to generate in real-world manufacturing. To address this, our framework employs self-supervised learning on unlabeled melt pool data using a Vision Transformer-based Masked Autoencoder (MAE) to produce highly representative embeddings. These fine-tuned embeddings are leveraged via transfer learning to train classifiers on a limited labeled dataset, enabling the effective identification of melt pool anomalies. We evaluate two classifiers: (1) a Vision Transformer (ViT) classifier utilizing the fine-tuned MAE Encoder's parameters and (2) the fine-tuned MAE Encoder combined with an MLP classifier head. Our framework achieves overall accuracy ranging from 95.44% to 99.17% and an average F1 score exceeding 80%, with the ViT Classifier slightly outperforming the MAE Encoder Classifier. This demonstrates the scalability and cost-effectiveness of our approach for automated quality control in DED, effectively detecting defects with minimal labeled data.
The NTK is a widely used tool in the theoretical analysis of deep learning, allowing us to look at supervised deep neural networks through the lenses of kernel regression. Recently, several works have investigated kernel models for self-supervised learning, hypothesizing that these also shed light on the behavior of wide neural networks by virtue of the NTK. However, it remains an open question to what extent this connection is mathematically sound -- it is a commonly encountered misbelief that the kernel behavior of wide neural networks emerges irrespective of the loss function it is trained on. In this paper, we bridge the gap between the NTK and self-supervised learning, focusing on two-layer neural networks trained under the Barlow Twins loss. We prove that the NTK of Barlow Twins indeed becomes constant as the width of the network approaches infinity. Our analysis technique is a bit different from previous works on the NTK and may be of independent interest. Overall, our work provides a first justification for the use of classic kernel theory to understand self-supervised learning of wide neural networks. Building on this result, we derive generalization error bounds for kernelized Barlow Twins and connect them to neural networks of finite width.
Masked Image Modeling (MIM) has emerged as a powerful self-supervised learning paradigm for visual representation learning, enabling models to acquire rich visual representations by predicting masked portions of images from their visible regions. While this approach has shown promising results, we hypothesize that its effectiveness may be limited by optimization challenges during early training stages, where models are expected to learn complex image distributions from partial observations before developing basic visual processing capabilities. To address this limitation, we propose a prototype-driven curriculum leagrning framework that structures the learning process to progress from prototypical examples to more complex variations in the dataset. Our approach introduces a temperature-based annealing scheme that gradually expands the training distribution, enabling more stable and efficient learning trajectories. Through extensive experiments on ImageNet-1K, we demonstrate that our curriculum learning strategy significantly improves both training efficiency and representation quality while requiring substantially fewer training epochs compared to standard Masked Auto-Encoding. Our findings suggest that carefully controlling the order of training examples plays a crucial role in self-supervised visual learning, providing a practical solution to the early-stage optimization challenges in MIM.
During surgical training, real-time feedback from trainers to trainees is important for preventing errors and enhancing long-term skill acquisition. Accurately predicting the effectiveness of this feedback, specifically whether it leads to a change in trainee behavior, is crucial for developing methods for improving surgical training and education. However, relying on human annotations to assess feedback effectiveness is laborious and prone to biases, underscoring the need for an automated, scalable, and objective method. Creating such an automated system poses challenges, as it requires an understanding of both the verbal feedback delivered by the trainer and the visual context of the real-time surgical scene. To address this, we propose a method that integrates information from transcribed verbal feedback and corresponding surgical video to predict feedback effectiveness. Our findings show that both transcribed feedback and surgical video are individually predictive of trainee behavior changes, and their combination achieves an AUROC of 0.70+/-0.02, improving prediction accuracy by up to 6.6%. Additionally, we introduce self-supervised fine-tuning as a strategy for enhancing surgical video representation learning, which is scalable and further enhances prediction performance. Our results demonstrate the potential of multi-modal learning to advance the automated assessment of surgical feedback.
Transformers and their variants have achieved great success in speech processing. However, their multi-head self-attention mechanism is computationally expensive. Therefore, one novel selective state space model, Mamba, has been proposed as an alternative. Building on its success in automatic speech recognition, we apply Mamba for spoofing attack detection. Mamba is well-suited for this task as it can capture the artifacts in spoofed speech signals by handling long-length sequences. However, Mamba's performance may suffer when it is trained with limited labeled data. To mitigate this, we propose combining a new structure of Mamba based on a dual-column architecture with self-supervised learning, using the pre-trained wav2vec 2.0 model. The experiments show that our proposed approach achieves competitive results and faster inference on the ASVspoof 2021 LA and DF datasets, and on the more challenging In-the-Wild dataset, it emerges as the strongest candidate for spoofing attack detection. The code has been publicly released in https://github.com/swagshaw/XLSR-Mamba.
Mobile grasping enhances manipulation efficiency by utilizing robots' mobility. This study aims to enable a commercial off-the-shelf robot for mobile grasping, requiring precise timing and pose adjustments. Self-supervised learning can develop a generalizable policy to adjust the robot's velocity and determine grasp position and orientation based on the target object's shape and pose. Due to mobile grasping's complexity, action primitivization and step-by-step learning are crucial to avoid data sparsity in learning from trial and error. This study simplifies mobile grasping into two grasp action primitives and a moving action primitive, which can be operated with limited degrees of freedom for the manipulator. This study introduces three fully convolutional neural network (FCN) models to predict static grasp primitive, dynamic grasp primitive, and residual moving velocity error from visual inputs. A two-stage grasp learning approach facilitates seamless FCN model learning. The ablation study demonstrated that the proposed method achieved the highest grasping accuracy and pick-and-place efficiency. Furthermore, randomizing object shapes and environments in the simulation effectively achieved generalizable mobile grasping.
Partial multi-view clustering (PVC) presents significant challenges practical research problem for data analysis in real-world applications, especially when some views of the data are partially missing. Existing clustering methods struggle to handle incomplete views effectively, leading to suboptimal clustering performance. In this paper, we propose a novel dual optimization framework based on contrastive learning, which aims to maximize the consistency of latent features in incomplete multi-view data and improve clustering performance through deep learning models. By combining a fine-tuned Vision Transformer and k-nearest neighbors (KNN), we fill in missing views and dynamically adjust view weights using self-supervised learning and meta-learning. Experimental results demonstrate that our framework outperforms state-of-the-art clustering models on the BDGP and HW datasets, particularly in handling complex and incomplete multi-view data.
Spatio-Temporal predictive Learning is a self-supervised learning paradigm that enables models to identify spatial and temporal patterns by predicting future frames based on past frames. Traditional methods, which use recurrent neural networks to capture temporal patterns, have proven their effectiveness but come with high system complexity and computational demand. Convolutions could offer a more efficient alternative but are limited by their characteristic of treating all previous frames equally, resulting in poor temporal characterization, and by their local receptive field, limiting the capacity to capture distant correlations among frames. In this paper, we propose STLight, a novel method for spatio-temporal learning that relies solely on channel-wise and depth-wise convolutions as learnable layers. STLight overcomes the limitations of traditional convolutional approaches by rearranging spatial and temporal dimensions together, using a single convolution to mix both types of features into a comprehensive spatio-temporal patch representation. This representation is then processed in a purely convolutional framework, capable of focusing simultaneously on the interaction among near and distant patches, and subsequently allowing for efficient reconstruction of the predicted frames. Our architecture achieves state-of-the-art performance on STL benchmarks across different datasets and settings, while significantly improving computational efficiency in terms of parameters and computational FLOPs. The code is publicly available