Large language models (LLMs) have demonstrated impressive reasoning ability in various language-based tasks. Despite many proposed reasoning methods aimed at enhancing performance in downstream tasks, two fundamental questions persist: Does reasoning genuinely support predictions, and how reliable is the quality of reasoning? In this paper, we propose a framework \textsc{SCORE} to analyze how well LLMs can reason. Specifically, we focus on self-contradictory reasoning, where reasoning does not support the prediction. We find that LLMs often contradict themselves when performing reasoning tasks that involve contextual information and commonsense. The model may miss evidence or use shortcuts, thereby exhibiting self-contradictory behaviors. We also employ the Point-of-View (POV) method, which probes models to generate reasoning from multiple perspectives, as a diagnostic tool for further analysis. We find that though LLMs may appear to perform well in one-perspective settings, they fail to stabilize such behavior in multi-perspectives settings. Even for correct predictions, the reasoning may be messy and incomplete, and LLMs can easily be led astray from good reasoning. \textsc{SCORE}'s results underscore the lack of robustness required for trustworthy reasoning and the urgency for further research to establish best practices for a comprehensive evaluation of reasoning beyond accuracy-based metrics.
The malicious applications of deep forgery, represented by face swapping, have introduced security threats such as misinformation dissemination and identity fraud. While some research has proposed the use of robust watermarking methods to trace the copyright of facial images for post-event traceability, these methods cannot effectively prevent the generation of forgeries at the source and curb their dissemination. To address this problem, we propose a novel comprehensive active defense mechanism that combines traceability and adversariality, called Dual Defense. Dual Defense invisibly embeds a single robust watermark within the target face to actively respond to sudden cases of malicious face swapping. It disrupts the output of the face swapping model while maintaining the integrity of watermark information throughout the entire dissemination process. This allows for watermark extraction at any stage of image tracking for traceability. Specifically, we introduce a watermark embedding network based on original-domain feature impersonation attack. This network learns robust adversarial features of target facial images and embeds watermarks, seeking a well-balanced trade-off between watermark invisibility, adversariality, and traceability through perceptual adversarial encoding strategies. Extensive experiments demonstrate that Dual Defense achieves optimal overall defense success rates and exhibits promising universality in anti-face swapping tasks and dataset generalization ability. It maintains impressive adversariality and traceability in both original and robust settings, surpassing current forgery defense methods that possess only one of these capabilities, including CMUA-Watermark, Anti-Forgery, FakeTagger, or PGD methods.
We propose BOSS, an approach that automatically learns to solve new long-horizon, complex, and meaningful tasks by growing a learned skill library with minimal supervision. Prior work in reinforcement learning require expert supervision, in the form of demonstrations or rich reward functions, to learn long-horizon tasks. Instead, our approach BOSS (BOotStrapping your own Skills) learns to accomplish new tasks by performing "skill bootstrapping," where an agent with a set of primitive skills interacts with the environment to practice new skills without receiving reward feedback for tasks outside of the initial skill set. This bootstrapping phase is guided by large language models (LLMs) that inform the agent of meaningful skills to chain together. Through this process, BOSS builds a wide range of complex and useful behaviors from a basic set of primitive skills. We demonstrate through experiments in realistic household environments that agents trained with our LLM-guided bootstrapping procedure outperform those trained with naive bootstrapping as well as prior unsupervised skill acquisition methods on zero-shot execution of unseen, long-horizon tasks in new environments. Website at clvrai.com/boss.
Deep learning models have a risk of utilizing spurious clues to make predictions, such as recognizing actions based on the background scene. This issue can severely degrade the open-set action recognition performance when the testing samples have different scene distributions from the training samples. To mitigate this problem, we propose a novel method, called Scene-debiasing Open-set Action Recognition (SOAR), which features an adversarial scene reconstruction module and an adaptive adversarial scene classification module. The former prevents the decoder from reconstructing the video background given video features, and thus helps reduce the background information in feature learning. The latter aims to confuse scene type classification given video features, with a specific emphasis on the action foreground, and helps to learn scene-invariant information. In addition, we design an experiment to quantify the scene bias. The results indicate that the current open-set action recognizers are biased toward the scene, and our proposed SOAR method better mitigates such bias. Furthermore, our extensive experiments demonstrate that our method outperforms state-of-the-art methods, and the ablation studies confirm the effectiveness of our proposed modules.
Among the remarkable emergent capabilities of large language models (LMs) is free-text rationalization; beyond a certain scale, large LMs are capable of generating seemingly useful rationalizations, which in turn, can dramatically enhance their performances on leaderboards. This phenomenon raises a question: can machine generated rationales also be useful for humans, especially when lay humans try to answer questions based on those machine rationales? We observe that human utility of existing rationales is far from satisfactory, and expensive to estimate with human studies. Existing metrics like task performance of the LM generating the rationales, or similarity between generated and gold rationales are not good indicators of their human utility. While we observe that certain properties of rationales like conciseness and novelty are correlated with their human utility, estimating them without human involvement is challenging. We show that, by estimating a rationale's helpfulness in answering similar unseen instances, we can measure its human utility to a better extent. We also translate this finding into an automated score, GEN-U, that we propose, which can help improve LMs' ability to generate rationales with better human utility, while maintaining most of its task performance. Lastly, we release all code and collected data with this project.
Machine learning (ML) based time series forecasting models often require and assume certain degrees of stationarity in the data when producing forecasts. However, in many real-world situations, the data distributions are not stationary and they can change over time while reducing the accuracy of the forecasting models, which in the ML literature is known as concept drift. Handling concept drift in forecasting is essential for many ML methods in use nowadays, however, the prior work only proposes methods to handle concept drift in the classification domain. To fill this gap, we explore concept drift handling methods in particular for Global Forecasting Models (GFM) which recently have gained popularity in the forecasting domain. We propose two new concept drift handling methods, namely: Error Contribution Weighting (ECW) and Gradient Descent Weighting (GDW), based on a continuous adaptive weighting concept. These methods use two forecasting models which are separately trained with the most recent series and all series, and finally, the weighted average of the forecasts provided by the two models are considered as the final forecasts. Using LightGBM as the underlying base learner, in our evaluation on three simulated datasets, the proposed models achieve significantly higher accuracy than a set of statistical benchmarks and LightGBM baselines across four evaluation metrics.
This study investigates clustered federated learning (FL), one of the formulations of FL with non-i.i.d. data, where the devices are partitioned into clusters and each cluster optimally fits its data with a localized model. We propose a novel clustered FL framework, which applies a nonconvex penalty to pairwise differences of parameters. This framework can automatically identify clusters without a priori knowledge of the number of clusters and the set of devices in each cluster. To implement the proposed framework, we develop a novel clustered FL method called FPFC. Advancing from the standard ADMM, our method is implemented in parallel, updates only a subset of devices at each communication round, and allows each participating device to perform a variable amount of work. This greatly reduces the communication cost while simultaneously preserving privacy, making it practical for FL. We also propose a new warmup strategy for hyperparameter tuning under FL settings and consider the asynchronous variant of FPFC (asyncFPFC). Theoretically, we provide convergence guarantees of FPFC for general nonconvex losses and establish the statistical convergence rate under a linear model with squared loss. Our extensive experiments demonstrate the advantages of FPFC over existing methods.
NLP models are susceptible to learning spurious biases (i.e., bugs) that work on some datasets but do not properly reflect the underlying task. Explanation-based model debugging aims to resolve spurious biases by showing human users explanations of model behavior, asking users to give feedback on the behavior, then using the feedback to update the model. While existing model debugging methods have shown promise, their prototype-level implementations provide limited practical utility. Thus, we propose XMD: the first open-source, end-to-end framework for explanation-based model debugging. Given task- or instance-level explanations, users can flexibly provide various forms of feedback via an intuitive, web-based UI. After receiving user feedback, XMD automatically updates the model in real time, by regularizing the model so that its explanations align with the user feedback. The new model can then be easily deployed into real-world applications via Hugging Face. Using XMD, we can improve the model's OOD performance on text classification tasks by up to 18%.
Benefiting from large-scale Pretrained Vision-Language Models (VL-PMs), the performance of Visual Question Answering (VQA) has started to approach human oracle performance. However, finetuning large-scale VL-PMs with limited data for VQA usually faces overfitting and poor generalization issues, leading to a lack of robustness. In this paper, we aim to improve the robustness of VQA systems (ie, the ability of the systems to defend against input variations and human-adversarial attacks) from the perspective of Information Bottleneck when finetuning VL-PMs for VQA. Generally, internal representations obtained by VL-PMs inevitably contain irrelevant and redundant information for the downstream VQA task, resulting in statistically spurious correlations and insensitivity to input variations. To encourage representations to converge to a minimal sufficient statistic in vision-language learning, we propose the Correlation Information Bottleneck (CIB) principle, which seeks a tradeoff between representation compression and redundancy by minimizing the mutual information (MI) between the inputs and internal representations while maximizing the MI between the outputs and the representations. Meanwhile, CIB measures the internal correlations among visual and linguistic inputs and representations by a symmetrized joint MI estimation. Extensive experiments on five VQA benchmarks of input robustness and two VQA benchmarks of human-adversarial robustness demonstrate the effectiveness and superiority of the proposed CIB in improving the robustness of VQA systems.
To automatically correct handwritten assignments, the traditional approach is to use an OCR model to recognize characters and compare them to answers. The OCR model easily gets confused on recognizing handwritten Chinese characters, and the textual information of the answers is missing during the model inference. However, teachers always have these answers in mind to review and correct assignments. In this paper, we focus on the Chinese cloze tests correction and propose a multimodal approach (named AiM). The encoded representations of answers interact with the visual information of students' handwriting. Instead of predicting 'right' or 'wrong', we perform the sequence labeling on the answer text to infer which answer character differs from the handwritten content in a fine-grained way. We take samples of OCR datasets as the positive samples for this task, and develop a negative sample augmentation method to scale up the training data. Experimental results show that AiM outperforms OCR-based methods by a large margin. Extensive studies demonstrate the effectiveness of our multimodal approach.