The recent performance leap of Large Language Models (LLMs) opens up new opportunities across numerous industrial applications and domains. However, erroneous generations, such as false predictions, misinformation, and hallucination made by LLMs, have also raised severe concerns for the trustworthiness of LLMs', especially in safety-, security- and reliability-sensitive scenarios, potentially hindering real-world adoptions. While uncertainty estimation has shown its potential for interpreting the prediction risks made by general machine learning (ML) models, little is known about whether and to what extent it can help explore an LLM's capabilities and counteract its undesired behavior. To bridge the gap, in this paper, we initiate an exploratory study on the risk assessment of LLMs from the lens of uncertainty. In particular, we experiment with twelve uncertainty estimation methods and four LLMs on four prominent natural language processing (NLP) tasks to investigate to what extent uncertainty estimation techniques could help characterize the prediction risks of LLMs. Our findings validate the effectiveness of uncertainty estimation for revealing LLMs' uncertain/non-factual predictions. In addition to general NLP tasks, we extensively conduct experiments with four LLMs for code generation on two datasets. We find that uncertainty estimation can potentially uncover buggy programs generated by LLMs. Insights from our study shed light on future design and development for reliable LLMs, facilitating further research toward enhancing the trustworthiness of LLMs.

Software Vulnerabilities (SVs) are increasing in complexity and scale, posing great security risks to many software systems. Given the limited resources in practice, SV assessment and prioritization help practitioners devise optimal SV mitigation plans based on various SV characteristics. The surge in SV data sources and data-driven techniques such as Machine Learning and Deep Learning have taken SV assessment and prioritization to the next level. Our survey provides a taxonomy of the past research efforts and highlights the best practices for data-driven SV assessment and prioritization. We also discuss the current limitations and propose potential solutions to address such issues.

Production recommendation systems rely on embedding methods to represent various features. An impeding challenge in practice is that the large embedding matrix incurs substantial memory footprint in serving as the number of features grows over time. We propose a similarity-aware embedding matrix compression method called Saec to address this challenge. Saec clusters similar features within a field to reduce the embedding matrix size. Saec also adopts a fast clustering optimization based on feature frequency to drastically improve clustering time. We implement and evaluate Saec on Numerous, the production distributed machine learning system in Tencent, with 10-day worth of feature data from QQ mobile browser. Testbed experiments show that Saec reduces the number of embedding vectors by two orders of magnitude, compresses the embedding size by ~27x, and delivers the same AUC and log loss performance.

This paper aims to classify a single PCG recording as normal or abnormal for computer-aided diagnosis. The proposed framework for this challenge has four steps: preprocessing, feature extraction, training and validation. In the preprocessing step, a recording is segmented into four states, i.e., the first heart sound, systolic interval, the second heart sound, and diastolic interval by the Springer Segmentation algorithm. In the feature extraction step, the authors extract 324 features from multi-domains to perform classification. A back propagation neural network is used as predication model. The optimal threshold for distinguishing normal and abnormal is determined by the statistics of model output for both normal and abnormal. The performance of the proposed predictor tested by the six training sets is sensitivity 0.812 and specificity 0.860 (overall accuracy is 0.836). However, the performance reduces to sensitivity 0.807 and specificity 0.829 (overall accuracy is 0.818) for the hidden test set.