Recently, there have been efforts to encode the linguistic information of speech using a self-supervised framework for speech synthesis. However, predicting representations from surrounding representations can inadvertently entangle speaker information in the speech representation. This paper aims to remove speaker information by exploiting the structured nature of speech, composed of discrete units like phonemes with clear boundaries. A neural network predicts these boundaries, enabling variable-length pooling for event-based representation extraction instead of fixed-rate methods. The boundary predictor outputs a probability for the boundary between 0 and 1, making pooling soft. The model is trained to minimize the difference with the pooled representation of the data augmented by time-stretch and pitch-shift. To confirm that the learned representation includes contents information but is independent of speaker information, the model was evaluated with libri-light's phonetic ABX task and SUPERB's speaker identification task.
The field of visually rich document understanding (VRDU) aims to solve a multitude of well-researched NLP tasks in a multi-modal domain. Several datasets exist for research on specific tasks of VRDU such as document classification (DC), key entity extraction (KEE), entity linking, visual question answering (VQA), inter alia. These datasets cover documents like invoices and receipts with sparse annotations such that they support one or two co-related tasks (e.g., entity extraction and entity linking). Unfortunately, only focusing on a single specific of documents or task is not representative of how documents often need to be processed in the wild - where variety in style and requirements is expected. In this paper, we introduce BuDDIE (Business Document Dataset for Information Extraction), the first multi-task dataset of 1,665 real-world business documents that contains rich and dense annotations for DC, KEE, and VQA. Our dataset consists of publicly available business entity documents from US state government websites. The documents are structured and vary in their style and layout across states and types (e.g., forms, certificates, reports, etc.). We provide data variety and quality metrics for BuDDIE as well as a series of baselines for each task. Our baselines cover traditional textual, multi-modal, and large language model approaches to VRDU.
An important use case of next-generation wireless systems is device-edge co-inference, where a semantic task is partitioned between a device and an edge server. The device carries out data collection and partial processing of the data, while the remote server completes the given task based on information received from the device. It is often required that processing and communication be run as efficiently as possible at the device, while more computing resources are available at the edge. To address such scenarios, we introduce a new system solution, termed neuromorphic wireless device-edge co-inference. According to it, the device runs sensing, processing, and communication units using neuromorphic hardware, while the server employs conventional radio and computing technologies. The proposed system is designed using a transmitter-centric information-theoretic criterion that targets a reduction of the communication overhead, while retaining the most relevant information for the end-to-end semantic task of interest. Numerical results on standard data sets validate the proposed architecture, and a preliminary testbed realization is reported.
Existing methods often adjust representations adaptively only after aggregating user behavior sequences. This coarse-grained approach to re-weighting the entire user sequence hampers the model's ability to accurately model the user interest migration across different scenarios. To enhance the model's capacity to capture user interests from historical behavior sequences in each scenario, we develop a ranking framework named the Scenario-Adaptive Fine-Grained Personalization Network (SFPNet), which designs a kind of fine-grained method for multi-scenario personalized recommendations. Specifically, SFPNet comprises a series of blocks named as Scenario-Tailoring Block, stacked sequentially. Each block initially deploys a parameter personalization unit to integrate scenario information at a coarse-grained level by redefining fundamental features. Subsequently, we consolidate scenario-adaptively adjusted feature representations to serve as context information. By employing residual connection, we incorporate this context into the representation of each historical behavior, allowing for context-aware fine-grained customization of the behavior representations at the scenario-level, which in turn supports scenario-aware user interest modeling.
Evaluating the reliability of news sources is a routine task for journalists and organizations committed to acquiring and disseminating accurate information. Recent research has shown that predicting sources' reliability represents an important first-prior step in addressing additional challenges such as fake news detection and fact-checking. In this paper, we introduce a novel approach for source reliability estimation that leverages reinforcement learning strategies for estimating the reliability degree of news sources. Contrary to previous research, our proposed approach models the problem as the estimation of a reliability degree, and not a reliability label, based on how all the news media sources interact with each other on the Web. We validated the effectiveness of our method on a news media reliability dataset that is an order of magnitude larger than comparable existing datasets. Results show that the estimated reliability degrees strongly correlates with journalists-provided scores (Spearman=0.80) and can effectively predict reliability labels (macro-avg. F$_1$ score=81.05). We release our implementation and dataset, aiming to provide a valuable resource for the NLP community working on information verification.
Effectively discerning spatial-spectral dependencies in HSI denoising is crucial, but prevailing methods using convolution or transformers still face computational efficiency limitations. Recently, the emerging Selective State Space Model(Mamba) has risen with its nearly linear computational complexity in processing natural language sequences, which inspired us to explore its potential in handling long spectral sequences. In this paper, we propose HSIDMamba(HSDM), tailored to exploit the linear complexity for effectively capturing spatial-spectral dependencies in HSI denoising. In particular, HSDM comprises multiple Hyperspectral Continuous Scan Blocks, incorporating BCSM(Bidirectional Continuous Scanning Mechanism), scale residual, and spectral attention mechanisms to enhance the capture of long-range and local spatial-spectral information. BCSM strengthens spatial-spectral interactions by linking forward and backward scans and enhancing information from eight directions through SSM, significantly enhancing the perceptual capability of HSDM and improving denoising performance more effectively. Extensive evaluations against HSI denoising benchmarks validate the superior performance of HSDM, achieving state-of-the-art results in performance and surpassing the efficiency of the latest transformer architectures by $30\%$.
Privacy-preserving analytics is designed to protect valuable assets. A common service provision involves the input data from the client and the model on the analyst's side. The importance of the privacy preservation is fuelled by legal obligations and intellectual property concerns. We explore the use case of a model owner providing an analytic service on customer's private data. No information about the data shall be revealed to the analyst and no information about the model shall be leaked to the customer. Current methods involve costs: accuracy deterioration and computational complexity. The complexity, in turn, results in a longer processing time, increased requirement on computing resources, and involves data communication between the client and the server. In order to deploy such service architecture, we need to evaluate the optimal setting that fits the constraints. And that is what this paper addresses. In this work, we enhance an attack detection system based on Convolutional Neural Networks with privacy-preserving technology based on PriMIA framework that is initially designed for medical data.
Research on diffusion model-based video generation has advanced rapidly. However, limitations in object fidelity and generation length hinder its practical applications. Additionally, specific domains like animated wallpapers require seamless looping, where the first and last frames of the video match seamlessly. To address these challenges, this paper proposes LoopAnimate, a novel method for generating videos with consistent start and end frames. To enhance object fidelity, we introduce a framework that decouples multi-level image appearance and textual semantic information. Building upon an image-to-image diffusion model, our approach incorporates both pixel-level and feature-level information from the input image, injecting image appearance and textual semantic embeddings at different positions of the diffusion model. Existing UNet-based video generation models require to input the entire videos during training to encode temporal and positional information at once. However, due to limitations in GPU memory, the number of frames is typically restricted to 16. To address this, this paper proposes a three-stage training strategy with progressively increasing frame numbers and reducing fine-tuning modules. Additionally, we introduce the Temporal E nhanced Motion Module(TEMM) to extend the capacity for encoding temporal and positional information up to 36 frames. The proposed LoopAnimate, which for the first time extends the single-pass generation length of UNet-based video generation models to 35 frames while maintaining high-quality video generation. Experiments demonstrate that LoopAnimate achieves state-of-the-art performance in both objective metrics, such as fidelity and temporal consistency, and subjective evaluation results.
Physical Reservoir Computing (PRC) is a recently developed variant of Neuromorphic Computing, where a pertinent physical system effectively projects information encoded in the input signal into a higher-dimensional space. While various physical hardware has demonstrated promising results for Reservoir Computing (RC), systems allowing tunability of their dynamical regimes have not received much attention regarding how to optimize relevant system parameters. In this work we employ hybrid photonic-electronic (HPE) system offering both parallelism inherent to light propagation, and electronic memory and programmable feedback allowing to induce nonlinear dynamics and tunable encoding of the photonic signal to realize HPE-RC. Specifically, we experimentally and theoretically analyze performance of integrated silicon photonic on-chip Mach-Zehnder interferometer and ring resonators with heaters acting as programmable phase modulators, controlled by detector and the feedback unit capable of realizing complex temporal dynamics of the photonic signal. Furthermore, we present an algorithm capable of predicting optimal parameters for RC by analyzing the corresponding Lyapunov exponent of the output signal and mutual information of reservoir nodes. By implementing the derived optimal parameters, we demonstrate that the corresponding resulting error of RC can be lowered by several orders of magnitude compared to a reservoir operating with randomly chosen set of parameters.
This paper introduces VLAP, a novel approach that bridges pretrained vision models and large language models (LLMs) to make frozen LLMs understand the visual world. VLAP transforms the embedding space of pretrained vision models into the LLMs' word embedding space using a single linear layer for efficient and general-purpose visual and language understanding. Specifically, we harness well-established word embeddings to bridge two modality embedding spaces. The visual and text representations are simultaneously assigned to a set of word embeddings within pretrained LLMs by formulating the assigning procedure as an optimal transport problem. We predict the assignment of one modality from the representation of another modality data, enforcing consistent assignments for paired multimodal data. This allows vision and language representations to contain the same information, grounding the frozen LLMs' word embedding space in visual data. Moreover, a robust semantic taxonomy of LLMs can be preserved with visual data since the LLMs interpret and reason linguistic information from correlations between word embeddings. Experimental results show that VLAP achieves substantial improvements over the previous linear transformation-based approaches across a range of vision-language tasks, including image captioning, visual question answering, and cross-modal retrieval. We also demonstrate the learned visual representations hold a semantic taxonomy of LLMs, making visual semantic arithmetic possible.