We propose Knowledge Boundary Discovery (KBD), a reinforcement learning based framework to explore the knowledge boundaries of the Large Language Models (LLMs). We define the knowledge boundary by automatically generating two types of questions: (i) those the LLM can confidently answer within-knowledge boundary and (ii) those it cannot beyond-knowledge boundary. Iteratively exploring and exploiting the LLM's responses to find its knowledge boundaries is challenging because of the hallucination phenomenon. To find the knowledge boundaries of an LLM, the agent interacts with the LLM under the modeling of exploring a partially observable environment. The agent generates a progressive question as the action, adopts an entropy reduction as the reward, receives the LLM's response as the observation and updates its belief states. We demonstrate that the KBD detects knowledge boundaries of LLMs by automatically finding a set of non-trivial answerable and unanswerable questions. We validate the KBD by comparing its generated knowledge boundaries with manually crafted LLM benchmark datasets. Experiments show that our KBD-generated question set is comparable to the human-generated datasets. Our approach paves a new way to evaluate LLMs.

Trusted Execution Environment(TEE) is a primary means for confidential computing. However, at the moment the RISC-V platform is limited for confidential computing because current RISC-V TEEs either lack scalability or compatibility. The reason for this dilemma in scalability and compatibility is that the standard isolation primitive on RISC-V,Physical Memory Protection(PMP), is not scalable. Meanwhile, previous enclave designs depend on theRich Execution Environment(REE) for OS functionalities, which increases domain switch frequency and enlarges the attack surface of the TEE. In this work, we propose Coffer, a scalable and efficient software-based TEE for the standard RISC-V platform. Coffer includes two core techniques:Logical PMP(LPMP) andEnclave Modules(EModules) to address the issues mentioned above. LPMP is a secure and efficient framework for PMP virtualization. It provides both scalability and hardware compatibility to Coffer. EModules are dynamically assembled lightweight libraries to provide enclaves with OS functionalities. The EModules provide Coffer with software compatibility and reduce theTrusted Computing Base(TCB) size of the enclaves. We implement and evaluate Coffer on commercially available RISC-V devices. The evaluation results show that Coffer can support 2, 000+ concurrent enclaves with negligible performance overhead. Particularly, LPMP supports enclave execution under heavy memory fragmentation with little performance overhead.

Multi-instance multi-label classification (MIML) is a fundamental task in machine learning, where each data sample comprises a bag containing several instances and multiple binary labels. Despite its wide applications, the data collection process involves matching multiple instances and labels, typically resulting in high annotation costs. In this paper, we study a novel yet practical crowdsourced multi-instance multi-label classification (CMIML) setup, where labels are collected from multiple crowd sources. To address this problem, we first propose a novel data generation process for CMIML, i.e., cross-label transition, where cross-label annotation error is more likely to appear rather than previous single-label transition assumption, due to the inherent similarity of localized instances from different classes. Then, we formally define the cross-label transition by cross-label transition matrices which are dependent across classes. Subsequently, we establish the first unbiased risk estimator for CMIML and further improve it through aggregation techniques, along with a rigorous generalization error bound. We also provide a practical implementation of cross-label transition matrix estimation. Comprehensive experiments on six benchmark datasets under various scenarios demonstrate that our algorithm outperforms the baselines by a large margin, validating its effectiveness in handling the CMIML problem.

Semantic Segmentation in the Foggy Scenes (SSFS) remains a difficult problem due to uncertainties caused by imperfect observations. Considering the success of domain adaptive semantic segmentation in the clear scenes, we believe it is reasonable to transfer the knowledge from the clear images to the foggy images. Different from the previous methods which mainly focus on narrowing the domain gap caused by fog, we try to transfer both the knowledge of fog factors and style factors between different domains to a "teacher" seg-mentor, thus the latter can generate better pseudo labels to supervise the student segmentor (main segmentor) to close the domain gap. Our method achieved better performance on ACDC and Foggy Zurich benchmark compared with state-of-the-art works.

ICESat-2 can obtain high-precision three-dimensional measurement information of targets and has a unique advantage in determining global elevation control points. However, due to the influence of the atmospheric environment, target characteristics, hardware equipment, and other factors, its elevation accuracy is not highly reliable, and not all data points can be used as control points. To obtain high-precision elevation control points from ICESat-2 data products, an extraction strategy that combines the accuracy and location requirements for control points was developed based on the ATL08 product. Multiple attribute parameters were incorporated into the extraction procedure, including terrain factor information, segment elevation information, cloud confidence flag, surface coverage data, topographic photon quantity, and photon height difference information. The extraction approach was then performed and analyzed using experimental data from the Hanzhong area in Shanxi province and the Songshan area in Henan province. In the experiments, the root-mean-square error (RMSE) of the extracted data points in two areas were both about 0.5 m and could reach 0.3 m after eliminating gross error points caused by surface changes and misclassification. The results suggest that the developed strategy takes into account the point location requirements of control points, can overcome the influence of complex terrain and ground objects and significantly improve the overall accuracy of obtained control points. Therefore, the proposed extraction strategy can support the establishment of the global elevation control point database and promote the application of ICESat-2 data in land surface surveying and mapping.