Image fusion is the process of combining images from different sensors into a single image that incorporates all relevant information. The majority of state-of-the-art image fusion techniques use deep learning methods to extract meaningful features; however, they primarily integrate local features without considering the image's broader context. To overcome this limitation, Transformer-based models have emerged as a promising solution, aiming to capture general context dependencies through attention mechanisms. Since there is no ground truth for image fusion, the loss functions are structured based on evaluation metrics, such as the structural similarity index measure (SSIM). By doing so, we create a bias towards the SSIM and, therefore, the input visual band image. The objective of this study is to propose a novel methodology for image fusion that mitigates the limitations associated with using evaluation metrics as loss functions. Our approach integrates a transformer-based multi-scale fusion strategy, which adeptly addresses both local and global context information. This integration not only refines the individual components of the image fusion process but also significantly enhances the overall efficacy of the method. Our proposed method follows a two-stage training approach, where an auto-encoder is initially trained to extract deep features at multiple scales at the first stage. For the second stage, we integrate our fusion block and change the loss function as mentioned. The multi-scale features are fused using a combination of Convolutional Neural Networks (CNNs) and Transformers. The CNNs are utilized to capture local features, while the Transformer handles the integration of general context features.
Extended Vision techniques are ubiquitous in physics. However, the data cubes steaming from such analysis often pose a challenge in their interpretation, due to the intrinsic difficulty in discerning the relevant information from the spectra composing the data cube. Furthermore, the huge dimensionality of data cube spectra poses a complex task in its statistical interpretation; nevertheless, this complexity contains a massive amount of statistical information that can be exploited in an unsupervised manner to outline some essential properties of the case study at hand, e.g.~it is possible to obtain an image segmentation via (deep) clustering of data-cube's spectra, performed in a suitably defined low-dimensional embedding space. To tackle this topic, we explore the possibility of applying unsupervised clustering methods in encoded space, i.e. perform deep clustering on the spectral properties of datacube pixels. A statistical dimensional reduction is performed by an ad hoc trained (Variational) AutoEncoder, in charge of mapping spectra into lower dimensional metric spaces, while the clustering process is performed by a (learnable) iterative K-Means clustering algorithm. We apply this technique to two different use cases, of different physical origins: a set of Macro mapping X-Ray Fluorescence (MA-XRF) synthetic data on pictorial artworks, and a dataset of simulated astrophysical observations.
Semantic communication has emerged as new paradigm shifts in 6G from the conventional syntax-oriented communications. Recently, the wireless broadcast technology has been introduced to support semantic communication system toward higher communication efficiency. Nevertheless, existing broadcast semantic communication systems target on general representation within one stage and fail to balance the inference accuracy among users. In this paper, the broadcast encoding process is decomposed into compression and fusion to improves communication efficiency with adaptation to tasks and channels.Particularly, we propose multiple task-channel-aware sub-encoders (TCE) and a channel-aware feature fusion sub-encoder (CFE) towards compression and fusion, respectively. In TCEs, multiple local-channel-aware attention blocks are employed to extract and compress task-relevant information for each user. In GFE, we introduce a global-channel-aware fine-tuning block to merge these compressed task-relevant signals into a compact broadcast signal. Notably, we retrieve the bottleneck in DeepBroadcast and leverage information bottleneck theory to further optimize the parameter tuning of TCEs and CFE.We substantiate our approach through experiments on a range of heterogeneous tasks across various channels with additive white Gaussian noise (AWGN) channel, Rayleigh fading channel, and Rician fading channel. Simulation results evidence that the proposed DeepBroadcast outperforms the state-of-the-art methods.
Machine learning models traditionally assume that training and test data are independently and identically distributed. However, in real-world applications, the test distribution often differs from training. This problem, known as out-of-distribution generalization, challenges conventional models. Invariant Risk Minimization (IRM) emerges as a solution, aiming to identify features invariant across different environments to enhance out-of-distribution robustness. However, IRM's complexity, particularly its bi-level optimization, has led to the development of various approximate methods. Our study investigates these approximate IRM techniques, employing the Expected Calibration Error (ECE) as a key metric. ECE, which measures the reliability of model prediction, serves as an indicator of whether models effectively capture environment-invariant features. Through a comparative analysis of datasets with distributional shifts, we observe that Information Bottleneck-based IRM, which condenses representational information, achieves a balance in improving ECE while preserving accuracy relatively. This finding is pivotal, as it demonstrates a feasible path to maintaining robustness without compromising accuracy. Nonetheless, our experiments also caution against over-regularization, which can diminish accuracy. This underscores the necessity for a systematic approach in evaluating out-of-distribution generalization metrics, one that beyond mere accuracy to address the nuanced interplay between accuracy and calibration.
Tactile sensing is significant for robotics since it can obtain physical contact information during manipulation. To capture multimodal contact information within a compact framework, we designed a novel sensor called ViTacTip, which seamlessly integrates both tactile and visual perception capabilities into a single, integrated sensor unit. ViTacTip features a transparent skin to capture fine features of objects during contact, which can be known as the see-through-skin mechanism. In the meantime, the biomimetic tips embedded in ViTacTip can amplify touch motions during tactile perception. For comparative analysis, we also fabricated a ViTac sensor devoid of biomimetic tips, as well as a TacTip sensor with opaque skin. Furthermore, we develop a Generative Adversarial Network (GAN)-based approach for modality switching between different perception modes, effectively alternating the emphasis between vision and tactile perception modes. We conducted a performance evaluation of the proposed sensor across three distinct tasks: i) grating identification, ii) pose regression, and iii) contact localization and force estimation. In the grating identification task, ViTacTip demonstrated an accuracy of 99.72%, surpassing TacTip, which achieved 94.60%. It also exhibited superior performance in both pose and force estimation tasks with the minimum error of 0.08mm and 0.03N, respectively, in contrast to ViTac's 0.12mm and 0.15N. Results indicate that ViTacTip outperforms single-modality sensors.
Large language models (LLMs) have revolutionized the field of natural language processing, extending their strong capabilities into multi-modal domains. Thus, it is vital to define proper and diversified metrics for the evaluation of LLMs. In this paper, we introduce matrix entropy, a novel metric rooted in information theory and geometry principles to quantify the data compression proficiency in LLMs. It reflects the model's ability to extract relevant information and eliminate unnecessary elements, thereby providing insight into the language model's intrinsic capability. Specifically, we demonstrate its applicability in both single-modal (language) and multi-modal settings. For language models, our findings reveal that the matrix entropy of representations follows a scaling law type reduction when the model scales up, serving as a complement to the traditional loss scaling law. For the multi-modal setting, we also propose an evaluation method based on matrix entropy for assessing alignment quality and we find that modern large multi-modal models exhibit great alignment performance.
Multi-image super-resolution (MISR) allows to increase the spatial resolution of a low-resolution (LR) acquisition by combining multiple images carrying complementary information in the form of sub-pixel offsets in the scene sampling, and can be significantly more effective than its single-image counterpart. Its main difficulty lies in accurately registering and fusing the multi-image information. Currently studied settings, such as burst photography, typically involve assumptions of small geometric disparity between the LR images and rely on optical flow for image registration. We study a MISR method that can increase the resolution of sets of images acquired with arbitrary, and potentially wildly different, camera positions and orientations, generalizing the currently studied MISR settings. Our proposed model, called EpiMISR, moves away from optical flow and explicitly uses the epipolar geometry of the acquisition process, together with transformer-based processing of radiance feature fields to substantially improve over state-of-the-art MISR methods in presence of large disparities in the LR images.
Multiview clustering (MVC) segregates data samples into meaningful clusters by synthesizing information across multiple views. Moreover, deep learning-based methods have demonstrated their strong feature learning capabilities in MVC scenarios. However, effectively generalizing feature representations while maintaining consistency is still an intractable problem. In addition, most existing deep clustering methods based on contrastive learning overlook the consistency of the clustering representations during the clustering process. In this paper, we show how the above problems can be overcome and propose a consistent enhancement-based deep MVC method via contrastive learning (CCEC). Specifically, semantic connection blocks are incorporated into a feature representation to preserve the consistent information among multiple views. Furthermore, the representation process for clustering is enhanced through spectral clustering, and the consistency across multiple views is improved. Experiments conducted on five datasets demonstrate the effectiveness and superiority of our method in comparison with the state-of-the-art (SOTA) methods. The code for this method can be accessed at https://anonymous.4open.science/r/CCEC-E84E/.
In this work, we leverage the intrinsic segmentation of language sequences and design a new positional encoding method called Bilevel Positional Encoding (BiPE). For each position, our BiPE blends an intra-segment encoding and an inter-segment encoding. The intra-segment encoding identifies the locations within a segment and helps the model capture the semantic information therein via absolute positional encoding. The inter-segment encoding specifies the segment index, models the relationships between segments, and aims to improve extrapolation capabilities via relative positional encoding. Theoretical analysis shows this disentanglement of positional information makes learning more effective. The empirical results also show that our BiPE has superior length extrapolation capabilities across a wide range of tasks in diverse text modalities.
Table question answering is a popular task that assesses a model's ability to understand and interact with structured data. However, the given table often does not contain sufficient information for answering the question, necessitating the integration of external knowledge. Existing methods either convert both the table and external knowledge into text, which neglects the structured nature of the table; or they embed queries for external sources in the interaction with the table, which complicates the process. In this paper, we propose a simple yet effective method to integrate external information in a given table. Our method first constructs an augmenting table containing the missing information and then generates a SQL query over the two tables to answer the question. Experiments show that our method outperforms strong baselines on three table QA benchmarks. Our code is publicly available at https://github.com/UCSB-NLP-Chang/Augment_tableQA.