We replace the multiplication and sigmoid function of the conventional recurrent gate with addition and ReLU activation. This mechanism is designed to maintain long-term memory for sequence processing but at a reduced computational cost, thereby opening up for more efficient execution or larger models on restricted hardware. Recurrent Neural Networks (RNNs) with gating mechanisms such as LSTM and GRU have been widely successful in learning from sequential data due to their ability to capture long-term dependencies. Conventionally, the update based on current inputs and the previous state history is each multiplied with dynamic weights and combined to compute the next state. However, multiplication can be computationally expensive, especially for certain hardware architectures or alternative arithmetic systems such as homomorphic encryption. It is demonstrated that the novel gating mechanism can capture long-term dependencies for a standard synthetic sequence learning task while significantly reducing computational costs such that execution time is reduced by half on CPU and by one-third under encryption. Experimental results on handwritten text recognition tasks furthermore show that the proposed architecture can be trained to achieve comparable accuracy to conventional GRU and LSTM baselines. The gating mechanism introduced in this paper may enable privacy-preserving AI applications operating under homomorphic encryption by avoiding the multiplication of encrypted variables. It can also support quantization in (unencrypted) plaintext applications, with the potential for substantial performance gains since the addition-based formulation can avoid the expansion to double precision often required for multiplication.
Automated medical report generation has become increasingly important in medical analysis. It can produce computer-aided diagnosis descriptions and thus significantly alleviate the doctors' work. Inspired by the huge success of neural machine translation and image captioning, various deep learning methods have been proposed for medical report generation. However, due to the inherent properties of medical data, including data imbalance and the length and correlation between report sequences, the generated reports by existing methods may exhibit linguistic fluency but lack adequate clinical accuracy. In this work, we propose an image-to-indicator hierarchical transformer (IIHT) framework for medical report generation. It consists of three modules, i.e., a classifier module, an indicator expansion module and a generator module. The classifier module first extracts image features from the input medical images and produces disease-related indicators with their corresponding states. The disease-related indicators are subsequently utilised as input for the indicator expansion module, incorporating the "data-text-data" strategy. The transformer-based generator then leverages these extracted features along with image features as auxiliary information to generate final reports. Furthermore, the proposed IIHT method is feasible for radiologists to modify disease indicators in real-world scenarios and integrate the operations into the indicator expansion module for fluent and accurate medical report generation. Extensive experiments and comparisons with state-of-the-art methods under various evaluation metrics demonstrate the great performance of the proposed method.
The proliferation of short video and live-streaming platforms has revolutionized how consumers engage in online shopping. Instead of browsing product pages, consumers are now turning to rich-content e-commerce, where they can purchase products through dynamic and interactive media like short videos and live streams. This emerging form of online shopping has introduced technical challenges, as products may be presented differently across various media domains. Therefore, a unified product representation is essential for achieving cross-domain product recognition to ensure an optimal user search experience and effective product recommendations. Despite the urgent industrial need for a unified cross-domain product representation, previous studies have predominantly focused only on product pages without taking into account short videos and live streams. To fill the gap in the rich-content e-commerce area, in this paper, we introduce a large-scale cRoss-dOmain Product Ecognition dataset, called ROPE. ROPE covers a wide range of product categories and contains over 180,000 products, corresponding to millions of short videos and live streams. It is the first dataset to cover product pages, short videos, and live streams simultaneously, providing the basis for establishing a unified product representation across different media domains. Furthermore, we propose a Cross-dOmain Product rEpresentation framework, namely COPE, which unifies product representations in different domains through multimodal learning including text and vision. Extensive experiments on downstream tasks demonstrate the effectiveness of COPE in learning a joint feature space for all product domains.
Current deep networks are very data-hungry and benefit from training on largescale datasets, which are often time-consuming to collect and annotate. By contrast, synthetic data can be generated infinitely using generative models such as DALL-E and diffusion models, with minimal effort and cost. In this paper, we present DatasetDM, a generic dataset generation model that can produce diverse synthetic images and the corresponding high-quality perception annotations (e.g., segmentation masks, and depth). Our method builds upon the pre-trained diffusion model and extends text-guided image synthesis to perception data generation. We show that the rich latent code of the diffusion model can be effectively decoded as accurate perception annotations using a decoder module. Training the decoder only needs less than 1% (around 100 images) manually labeled images, enabling the generation of an infinitely large annotated dataset. Then these synthetic data can be used for training various perception models for downstream tasks. To showcase the power of the proposed approach, we generate datasets with rich dense pixel-wise labels for a wide range of downstream tasks, including semantic segmentation, instance segmentation, and depth estimation. Notably, it achieves 1) state-of-the-art results on semantic segmentation and instance segmentation; 2) significantly more robust on domain generalization than using the real data alone; and state-of-the-art results in zero-shot segmentation setting; and 3) flexibility for efficient application and novel task composition (e.g., image editing). The project website and code can be found at https://weijiawu.github.io/DatasetDM_page/ and https://github.com/showlab/DatasetDM, respectively
Text-based collaborative filtering (TCF) has become the mainstream approach for text and news recommendation, utilizing text encoders, also known as language models (LMs), to represent items. However, existing TCF models primarily focus on using small or medium-sized LMs. It remains uncertain what impact replacing the item encoder with one of the largest and most powerful LMs, such as the 175-billion parameter GPT-3 model, would have on recommendation performance. Can we expect unprecedented results? To this end, we conduct an extensive series of experiments aimed at exploring the performance limits of the TCF paradigm. Specifically, we increase the size of item encoders from one hundred million to one hundred billion to reveal the scaling limits of the TCF paradigm. We then examine whether these extremely large LMs could enable a universal item representation for the recommendation task. Furthermore, we compare the performance of the TCF paradigm utilizing the most powerful LMs to the currently dominant ID embedding-based paradigm and investigate the transferability of this TCF paradigm. Finally, we compare TCF with the recently popularized prompt-based recommendation using ChatGPT. Our research findings have not only yielded positive results but also uncovered some surprising and previously unknown negative outcomes, which can inspire deeper reflection and innovative thinking regarding text-based recommender systems. Codes and datasets will be released for further research.
Given a prefix (context), open-ended generation aims to decode texts that are coherent, which don't abruptly drift from previous topics, and informative, which don't suffer from undesired repetitions. In this paper, we propose Look-back, an improved decoding algorithm that leverages the Kullback-Leibler divergence to track the distribution distance between current and historical decoding steps. Thus Look-back can automatically predict potential repetitive phrase and topic drift, and remove tokens that may cause the failure modes, restricting the next token probability distribution within a plausible distance to the history. We perform decoding experiments on document continuation and story generation, and demonstrate that Look-back is able to generate more fluent and coherent text, outperforming other strong decoding methods significantly in both automatic and human evaluations.
Short text clustering is challenging since it takes imbalanced and noisy data as inputs. Existing approaches cannot solve this problem well, since (1) they are prone to obtain degenerate solutions especially on heavy imbalanced datasets, and (2) they are vulnerable to noises. To tackle the above issues, we propose a Robust Short Text Clustering (RSTC) model to improve robustness against imbalanced and noisy data. RSTC includes two modules, i.e., pseudo-label generation module and robust representation learning module. The former generates pseudo-labels to provide supervision for the later, which contributes to more robust representations and correctly separated clusters. To provide robustness against the imbalance in data, we propose self-adaptive optimal transport in the pseudo-label generation module. To improve robustness against the noise in data, we further introduce both class-wise and instance-wise contrastive learning in the robust representation learning module. Our empirical studies on eight short text clustering datasets demonstrate that RSTC significantly outperforms the state-of-the-art models. The code is available at: https://github.com/hmllmh/RSTC.
As large language models have demonstrated impressive performance in many domains, recent works have adopted language models (LMs) as controllers of visual modules for vision-and-language tasks. While existing work focuses on equipping LMs with visual understanding, we propose two novel interpretable/explainable visual programming frameworks for text-to-image (T2I) generation and evaluation. First, we introduce VPGen, an interpretable step-by-step T2I generation framework that decomposes T2I generation into three steps: object/count generation, layout generation, and image generation. We employ an LM to handle the first two steps (object/count generation and layout generation), by finetuning it on text-layout pairs. Our step-by-step T2I generation framework provides stronger spatial control than end-to-end models, the dominant approach for this task. Furthermore, we leverage the world knowledge of pretrained LMs, overcoming the limitation of previous layout-guided T2I works that can only handle predefined object classes. We demonstrate that our VPGen has improved control in counts/spatial relations/scales of objects than state-of-the-art T2I generation models. Second, we introduce VPEval, an interpretable and explainable evaluation framework for T2I generation based on visual programming. Unlike previous T2I evaluations with a single scoring model that is accurate in some skills but unreliable in others, VPEval produces evaluation programs that invoke a set of visual modules that are experts in different skills, and also provides visual+textual explanations of the evaluation results. Our analysis shows VPEval provides a more human-correlated evaluation for skill-specific and open-ended prompts than widely used single model-based evaluation. We hope our work encourages future progress on interpretable/explainable generation and evaluation for T2I models. Website: https://vp-t2i.github.io
This work explores capabilities of the pre-trained CLIP vision-language model to identify satellite images affected by clouds. Several approaches to using the model to perform cloud presence detection are proposed and evaluated, including a purely zero-shot operation with text prompts and several fine-tuning approaches. Furthermore, the transferability of the methods across different datasets and sensor types (Sentinel-2 and Landsat-8) is tested. The results that CLIP can achieve non-trivial performance on the cloud presence detection task with apparent capability to generalise across sensing modalities and sensing bands. It is also found that a low-cost fine-tuning stage leads to a strong increase in true negative rate. The results demonstrate that the representations learned by the CLIP model can be useful for satellite image processing tasks involving clouds.
Large Language Models work quite well with general-purpose data and many tasks in Natural Language Processing. However, they show several limitations when used for a task such as domain-specific abstractive text summarization. This paper identifies three of those limitations as research problems in the context of abstractive text summarization: 1) Quadratic complexity of transformer-based models with respect to the input text length; 2) Model Hallucination, which is a model's ability to generate factually incorrect text; and 3) Domain Shift, which happens when the distribution of the model's training and test corpus is not the same. Along with a discussion of the open research questions, this paper also provides an assessment of existing state-of-the-art techniques relevant to domain-specific text summarization to address the research gaps.