Integrating domain-specific operators into processor cores is essential for performance scaling. However, multi-operand operators often face a semantic gap with conventional ISAs, which are limited in operand capacity and scheduling flexibility. This paper presents SMIX, a schedulable instruction set extension interface for multi-operand operators. SMIX decouples execution into three stages: out-of-order input filling, computation, and out-of-order result picking. By employing explicit encoding and counter-based dependency management, SMIX enables both efficient static scheduling by compilers and dynamic out-of-order execution in hardware. Experimental results demonstrate the high schedulability of SMIX, where static scheduling provides an average 12% performance gain on the Rocket core and dynamic out-of-order scheduling contributes an additional 9.2% speedup on the BOOM core, all while maintaining minimal hardware overhead.

Quantum computing has enabled precise simulation in gene regulatory network (GRN) prediction by capturing regulatory relationships at a microscopic level. However, current quantum approaches face significant limitations due to high noise levels and prohibitive training costs, making them challenging to implement effectively. To address these issues, we introduce Q-TetoFormer (Quantum-Tuned Extraction and Transformation Orchestrator Former), a novel framework that merges quantum computing with Transformer architecture to enhance GRN prediction accuracy and efficiency.Q-TetoFormer begins by preprocessing gene data, feeding it through a neural network-embedded quantum circuit. The circuit’s output then flows into TetoFormer, a Transformer-based module that leverages attention mechanisms to reduce quantum noise and enrich feature extraction. The output from TetoFormer subsequently optimizes the quantum circuit, creating a feedback loop that improves model performance iteratively. Extensive experiments on five GRN prediction benchmarks demonstrate that Q-TetoFormer not only achieves superior prediction accuracy over state-of-the-art methods but also significantly reduces training overhead, enhancing its practical utility.

Transactions are used to maintain the data integrity of databases, and have become an indispensable feature in modern Database Management Systems (DBMSs). Despite extensive efforts in testing DBMSs and verifying transaction processing mechanisms, isolation bugs still exist in widely-used DBMSs when these DBMSs violate their claimed transaction isolation levels. Isolation bugs can cause severe consequences, e.g., incorrect query results and database states. In this paper, we propose a novel transaction testing approach, Transaction oracle construction (Troc), to automatically detect isolation bugs in DBMSs. The core idea of Troc is to decouple a transaction into independent statements, and execute them on their own database views, which are constructed under the guidance of the claimed transaction isolation level. Any divergence between the actual transaction execution and the independent statement execution indicates an isolation bug. We implement and evaluate Troc on three widely-used DBMSs, i.e., MySQL, MariaDB, and TiDB. We have detected 5 previously-unknown isolation bugs in the latest versions of these DBMSs.

The success of batch mode active learning (BMAL) methods lies in selecting both representative and uncertain samples. Representative samples quickly capture the global structure of the whole dataset, while the uncertain ones refine the decision boundary. There are two principles, namely the direct approach and the screening approach, to make a trade-off between representativeness and uncertainty. Although widely used in literature, little is known about the relationship between these two principles. In this paper, we discover that the two approaches both have shortcomings in the initial stage of BMAL. To alleviate the shortcomings, we bound the certainty scores of unlabeled samples from below and directly combine this lower-bounded certainty with representativeness in the objective function. Additionally, we show that the two aforementioned approaches are mathematically equivalent to two special cases of our approach. To the best of our knowledge, this is the first work that tries to generalize the direct and screening approaches. The objective function is then solved by super-modularity optimization. Extensive experiments on fifteen datasets indicate that our method has significantly higher classification accuracy on testing data than the latest state-of-the-art BMAL methods, and also scales better even when the size of the unlabeled pool reaches 106.

Recognizing human actions in video clips has been an important topic in computer vision. Sufficient labeled data is one of the prerequisites for the good performance of action recognition algorithms. However, while abundant videos can be collected from the Internet, categorizing each video clip is tedious and even time-consuming. Active learning is one way to alleviate the labeling labor by allowing the classifier to choose the most informative unlabeled instances for manual annotation. Among various active learning algorithms, uncertainty sampling is arguably the most widely-used strategy. Conventional uncertainty sampling strategies such as entropy-based methods are usually tested under accuracy. However, in action recognition Average Precision (AP) is an acknowledged evaluation metric, which is somehow ignored in the active learning community. It is defined as the area under the precision-recall curve. In this paper, we propose a novel uncertainty sampling algorithm for action recognition using expected AP. We conduct experiments on three real-world action recognition datasets and show that our algorithm outperforms other uncertainty-based active learning algorithms.

Active learning is a machine learning technique that trains a classifier after selecting a subset from an unlabeled dataset for labeling and using the selected data for training. Recently, batch mode active learning, which selects a batch of samples to label in parallel, has attracted a lot of attention. Its challenge lies in the choice of criteria used for guiding the search of the optimal batch. In this paper, we propose a novel approach to selecting the optimal batch of queries by minimizing the α-relative Pearson divergence (RPE) between the labeled and the original datasets. This particular divergence is chosen since it can distinguish the optimal batch more easily than other measures especially when available candidates are similar. The proposed objective is a min-max optimization problem, and it is difficult to solve due to the involvement of both minimization and maximization. We find that the objective has an equivalent convex form, and thus a global optimal solution can be obtained. Then the subgradient method can be applied to solve the simplified convex problem. Our empirical studies on UCI datasets demonstrate the effectiveness of the proposed approach compared with the state-of-the-art batch mode active learning methods.

In classification tasks, labeled data is a necessity but sometimes difficult or expensive to obtain. On the contrary, unlabeled data is usually abundant. Recently, different active learning algorithms are proposed to alleviate this issue by selecting the most informative data points to label. One family of active learning methods comes from Optimum Experimental Design (OED) in statistics. Instead of selecting data points one by one iteratively, OED-based approaches select data in a one-shot manner, that is, a fixed-sized subset is selected from the unlabeled dataset for manually labeling. These methods usually use kernels to represent pair-wise similarities between different data points. It is well known that choosing optimal kernel types (e.g. Gaussian kernel) and kernel parameters (e.g. kernel width) is tricky, and a common way to resolve it is by Multiple Kernel Learning (MKL), i.e., to construct a few candidate kernels and merge them to form a consensus kernel. There would be different ways to combine multiple kernels, one of which, called the the globalised approach is to assign a weight to each candidate kernel. In practice different data points in the same candidate kernel may not have the same contribution in the consensus kernel, this requires assigning different weights to different data points in the same candidate kernel, leading to the localized approach. In this paper, we introduce MKL to OED-based active learning, specifically we propose globalised and localized multiple kernel active learning methods, respectively. Our experiments on six benchmark datasets demonstrate that the proposed methods have better performance than existing OED-based active learning methods.