Image-to-recipe retrieval is a challenging vision-to-language task of significant practical value. The main challenge of the task lies in the ultra-high redundancy in the long recipe and the large variation reflected in both food item combination and food item appearance. A de-facto idea to address this task is to learn a shared feature embedding space in which a food image is aligned better to its paired recipe than other recipes. However, such supervised global matching is prone to supervision collapse, i.e., only partial information that is necessary for distinguishing training pairs can be identified, while other information that is potentially useful in generalization could be lost. To mitigate such a problem, we propose a mask-augmentation-based local matching network (MALM), where an image-text matching module and a masked self-distillation module benefit each other mutually to learn generalizable cross-modality representations. On one hand, we perform local matching between the tokenized representations of image and text to locate fine-grained cross-modality correspondence explicitly. We involve representations of masked image patches in this process to alleviate overfitting resulting from local matching especially when some food items are underrepresented. On the other hand, predicting the hidden representations of the masked patches through self-distillation helps to learn general-purpose image representations that are expected to generalize better. And the multi-task nature of the model enables the representations of masked patches to be text-aware and thus facilitates the lost information reconstruction. Experimental results on Recipe1M dataset show our method can clearly outperform state-of-the-art (SOTA) methods. Our code will be available at https://github.com/MyFoodChoice/MALM_Mask_Augmentation_based_Local_Matching-_for-_Food_Recipe_Retrieval
Automatic radiology report generation is challenging as medical images or reports are usually similar to each other due to the common content of anatomy. This makes a model hard to capture the uniqueness of individual images and is prone to producing undesired generic or mismatched reports. This situation calls for learning more discriminative features that could capture even fine-grained mismatches between images and reports. To achieve this, this paper proposes a novel framework to learn discriminative image and report features by distinguishing them from their closest peers, i.e., hard negatives. Especially, to attain more discriminative features, we gradually raise the difficulty of such a learning task by creating increasingly hard negative reports for each image in the feature space during training, respectively. By treating the increasingly hard negatives as auxiliary variables, we formulate this process as a min-max alternating optimisation problem. At each iteration, conditioned on a given set of hard negative reports, image and report features are learned as usual by minimising the loss functions related to report generation. After that, a new set of harder negative reports will be created by maximising a loss reflecting image-report alignment. By solving this optimisation, we attain a model that can generate more specific and accurate reports. It is noteworthy that our framework enhances discriminative feature learning without introducing extra network weights. Also, in contrast to the existing way of generating hard negatives, our framework extends beyond the granularity of the dataset by generating harder samples out of the training set. Experimental study on benchmark datasets verifies the efficacy of our framework and shows that it can serve as a plug-in to readily improve existing medical report generation models.
Large Language Models (LLMs) have recently demonstrated exceptional performance in various Natural Language Processing (NLP) tasks. They have also shown the ability to perform chain-of-thought (CoT) reasoning to solve complex problems. Recent studies have explored CoT reasoning in complex multimodal scenarios, such as the science question answering task, by fine-tuning multimodal models with high-quality human-annotated CoT rationales. However, collecting high-quality COT rationales is usually time-consuming and costly. Besides, the annotated rationales are hardly accurate due to the redundant information involved or the essential information missed. To address these issues, we propose a novel method termed \emph{T-SciQ} that aims at teaching science question answering with LLM signals. The T-SciQ approach generates high-quality CoT rationales as teaching signals and is advanced to train much smaller models to perform CoT reasoning in complex modalities. Additionally, we introduce a novel data mixing strategy to produce more effective teaching data samples for simple and complex science question answer problems. Extensive experimental results show that our T-SciQ method achieves a new state-of-the-art performance on the ScienceQA benchmark, with an accuracy of 96.18%. Moreover, our approach outperforms the most powerful fine-tuned baseline by 4.5%.
This paper proposes a new depression detection system based on LLMs that is both interpretable and interactive. It not only provides a diagnosis, but also diagnostic evidence and personalized recommendations based on natural language dialogue with the user. We address challenges such as the processing of large amounts of text and integrate professional diagnostic criteria. Our system outperforms traditional methods across various settings and is demonstrated through case studies.
Existing MWP solvers employ sequence or binary tree to present the solution expression and decode it from given problem description. However, such structures fail to handle the identical variants derived via mathematical manipulation, e.g., $(a_1+a_2)*a_3$ and $a_1*a_3+a_2*a_3$ are for the same problem but formulating different expression sequences and trees, which would raise two issues in MWP solving: 1) different output solutions for the same input problem, making the model hard to learn the mapping function between input and output spaces, and 2) difficulty of evaluating solution expression that indicates wrong between the above examples. To address these issues, we first introduce a unified tree structure to present expression, where the elements are permutable and identical for all the expression variants. We then propose a novel non-autoregressive solver, dubbed MWP-NAS, to parse the problem and reason the solution expression based on the unified tree. For the second issue, to handle the variants in evaluation, we propose to match the unified tree and design a path-based metric to evaluate the partial accuracy of expression. Extensive experiments have been conducted on Math23K and MAWPS, and the results demonstrate the effectiveness of the proposed MWP-NAS. The codes and checkpoints are available at: https://github.com/mengqunhan/MWP-NAS
Large language models (LLMs) have recently been shown to deliver impressive performance in various NLP tasks. To tackle multi-step reasoning tasks, few-shot chain-of-thought (CoT) prompting includes a few manually crafted step-by-step reasoning demonstrations which enable LLMs to explicitly generate reasoning steps and improve their reasoning task accuracy. To eliminate the manual effort, Zero-shot-CoT concatenates the target problem statement with "Let's think step by step" as an input prompt to LLMs. Despite the success of Zero-shot-CoT, it still suffers from three pitfalls: calculation errors, missing-step errors, and semantic misunderstanding errors. To address the missing-step errors, we propose Plan-and-Solve (PS) Prompting. It consists of two components: first, devising a plan to divide the entire task into smaller subtasks, and then carrying out the subtasks according to the plan. To address the calculation errors and improve the quality of generated reasoning steps, we extend PS prompting with more detailed instructions and derive PS+ prompting. We evaluate our proposed prompting strategy on ten datasets across three reasoning problems. The experimental results over GPT-3 show that our proposed zero-shot prompting consistently outperforms Zero-shot-CoT across all datasets by a large margin, is comparable to or exceeds Zero-shot-Program-of-Thought Prompting, and has comparable performance with 8-shot CoT prompting on the math reasoning problem. The code can be found at https://github.com/AGI-Edgerunners/Plan-and-Solve-Prompting.
Visual representation based on covariance matrix has demonstrates its efficacy for image classification by characterising the pairwise correlation of different channels in convolutional feature maps. However, pairwise correlation will become misleading once there is another channel correlating with both channels of interest, resulting in the ``confounding'' effect. For this case, ``partial correlation'' which removes the confounding effect shall be estimated instead. Nevertheless, reliably estimating partial correlation requires to solve a symmetric positive definite matrix optimisation, known as sparse inverse covariance estimation (SICE). How to incorporate this process into CNN remains an open issue. In this work, we formulate SICE as a novel structured layer of CNN. To ensure end-to-end trainability, we develop an iterative method to solve the above matrix optimisation during forward and backward propagation steps. Our work obtains a partial correlation based deep visual representation and mitigates the small sample problem often encountered by covariance matrix estimation in CNN. Computationally, our model can be effectively trained with GPU and works well with a large number of channels of advanced CNNs. Experiments show the efficacy and superior classification performance of our deep visual representation compared to covariance matrix based counterparts.
Over the past few years, Federated Learning (FL) has become a popular distributed machine learning paradigm. FL involves a group of clients with decentralized data who collaborate to learn a common model under the coordination of a centralized server, with the goal of protecting clients' privacy by ensuring that local datasets never leave the clients and that the server only performs model aggregation. However, in realistic scenarios, the server may be able to collect a small amount of data that approximately mimics the population distribution and has stronger computational ability to perform the learning process. To address this, we focus on the hybrid FL framework in this paper. While previous hybrid FL work has shown that the alternative training of clients and server can increase convergence speed, it has focused on the scenario where clients fully participate and ignores the negative effect of partial participation. In this paper, we provide theoretical analysis of hybrid FL under clients' partial participation to validate that partial participation is the key constraint on convergence speed. We then propose a new algorithm called FedCLG, which investigates the two-fold role of the server in hybrid FL. Firstly, the server needs to process the training steps using its small amount of local datasets. Secondly, the server's calculated gradient needs to guide the participated clients' training and the server's aggregation. We validate our theoretical findings through numerical experiments, which show that our proposed method FedCLG outperforms state-of-the-art methods.