Large language models (LLMs) have demonstrated impressive capabilities in code generation, by leveraging retrieval-augmented generation (RAG) methods. However, the computational costs associated with LLM inference, particularly in terms of latency and energy consumption, have received limited attention in the security context. This paper introduces DrainCode, the first adversarial attack targeting the computational efficiency of RAG-based code generation systems. By strategically poisoning retrieval contexts through mutation-based approach, DrainCode forces LLMs to produce significantly longer outputs, thereby increasing GPU latency and energy consumption. We evaluate the effectiveness of DrainCode across multiple models. Our experiments show that DrainCode achieves up to a 85% increase in latency, a 49% increase in energy consumption, and more than a 3× increase in output length compared to the baseline. Furthermore, we demonstrate the generalizability of the attack across different prompting strategies and its effectiveness compared to different defenses. The results highlight DrainCode as a potential method for increasing the computational overhead of LLMs, making it useful for evaluating LLM security in resource-constrained environments. We provide code and data at https://github.com/DeepSoftwareAnalytics/DrainCode.

Cardiotocography (CTG) is essential for monitoring fetal health. Current deep learning applications in this field face two challenges: 1) the scarcity of annotated CTG data; 2) the difficulty in simultaneously capturing the coupled dependencies within multivariate time-series CTG. To bridge this gap, we propose MCTG: a multimodal self-supervised contrastive learning framework. For positive samples, we integrate an augmentation method based on frequency shapelets. These short frequency-domain subsequences capture class-specific information, preserving the most discriminative components of CTG signals. Additionally, we construct a multimodal learning framework that integrates series-image modalities to obtain complementary information between different modalities. This approach compensates for the inherent limitations of single time series modality networks in terms of complex cross-variable dependencies. Specifically, we encode series into images through three-channel encoding (frequency, wavelet, and time) to capture global long-term features, cross-scale features, and temporal texture details. Concurrently, multi-scale convolution integrates each variable dimension of the time series into a single image, enhancing the model’s ability to capture dependency couplings between variables. Experimental results demonstrate that MCTG effectively learns crucial feature representations from CTG data and achieves state-of-the-art performance in fetal distress prediction. The codes are available at https://github.com/Ladyfish030/MCTG.

Smart contracts are one of the most successful applications of blockchain technology. In order to guarantee the security of smart contracts, researchers have successively introduced various testing methodologies, including static analysis, symbolic execution, and fuzzing, which contribute to a more rigorous and precise evaluation of smart contract vulnerabilities. Deep learning techniques have been widely applied in traditional software vulnerability detection, while the opposite is true in the field of smart contract testing. Consequently, we anticipate that deep learning can be similarly applied to enhance traditional smart contract vulnerability detection tools. However, there is a lack of empirical study on the performance of deep learning applied to smart contract testing. In order to explore how deep neural networks can help with testing tools on smart contracts, we construct a test framework based on SMARTEST. We manage to train deep learning language models using various neural networks including Transformer, GRU, RNN and test the symbolic execution tool SMARTEST framework with the application of these models on the CVE dataset. Upon analyzing the experimental results, we find that deep neural networks did not surpass traditional language models in enhancing smart contract testing. In terms of accuracy, the SMARTEST tool, which utilizes a statistical 3-gram language model, succeeded in detecting the greatest number of vulnerabilities. Specifically, the 3-gram model was able to identify 69.8% of vulnerabilities in the benchmark set within the first 5 seconds. Based on our experimental findings and thorough analysis, we outline the challenges faced in DNN-assisted smart contract testing and suggest potential directions for improvement.

Bias Temperature Instability (BTI) and Hot Carrier Injections (HCI) are two of the main effects that increase a transistor's threshold voltage and further cause performance degradations. These two wearout mechanisms affect all transistors, but are especially acute in the SRAM cells of first-level (L1) caches, which are frequently accessed and are critical for microprocessor performance. This work studies the cache lifetimes due to the combined effect of BTI and HCI for different cache configurations, including variation in cache size, associativity, cache line size, and the replacement algorithm. The effect of process variations is also considered. We analyze the reliability (failure probability) and performance (hit rate) of the L1 cache within a LEON3 microprocessor, while the LEON3 is running a set of benchmarks, and we provide essential insights on performance-reliability tradeoffs for cache designers.

We present a Bayesian approach to adapting parameters of a well-trained context-dependent, deep-neural-network, hidden Markov model (CD-DNN-HMM) to improve automatic speech recognition performance. Given an abundance of DNN parameters but with only a limited amount of data, the effectiveness of the adapted DNN model can often be compromised. We formulate maximum a posteriori (MAP) adaptation of parameters of a specially designed CD-DNN-HMM with an augmented linear hidden networks connected to the output tied states, or senones, and compare it to feature space MAP linear regression previously proposed. Experimental evidences on the 20,000-word open vocabulary Wall Street Journal task demonstrate the feasibility of the proposed framework. In supervised adaptation, the proposed MAP adaptation approach provides more than 10% relative error reduction and consistently outperforms the conventional transformation based methods. Furthermore, we present an initial attempt to generate hierarchical priors to improve adaptation efficiency and effectiveness with limited adaptation data by exploiting similarities among senones.

The Hadoop HDFS is a popular open-source distributed storage system, which serves as the foundation of many important big-data technologies. The performance of data replication is crucial to HDFS, since it accounts for a major portion of network traffic in the entire cluster. In this research, we propose to enable multicast-based replication, which is expected to use less network bandwidth than the native TCP-based pipelined replication method. We developed a congestion-controlled reliable multicast socket (the CCRMSocket) for HDFS and evaluated its performance with our multi-rack test platform. The experimental result shows that our multicast implementation can effectively save bandwidth and peacefully coexist with TCP traffic. We also developed a simulator (the HFlowSim) to further study the impact of multicast-based replication to a large-scale Hadoop system. The simulation result suggests that multicast-based replication can systematically improve a Hadoop system by accelerating the big jobs.

In recent years, GPU has evolved rapidly and exhibited great potential in accelerating scientific applications. Massive GPU-assisted HPC systems have been deployed. However, as a heterogeneous system, GPU-assisted HPC is harder to be programmed and utilized than conventional CPU-only system. Statistics of the Keene land system indicate that the effective utilization rate of computational resources is only about 40% when the system runs in normal condition with enough jobs in its queue. Our theoretical model shows that the lack of overlap between CPU/GPU computation is a major obstacle in the efficient utilization of heterogeneous system. In this paper, we evaluate the possibility of collocating CPU-only job with GPU-assisted job on the same node to increase overlap between CPU/GPU computation, thus achieving better utilization. Several performance compromising factors, such as resource isolation, CPU load, and GPU memory demands, are studied based on workload from popular MPI/CUDA benchmarks. The results indicate that, when those factors are managed properly, the collocated CPU-only job can efficiently scavenge the underutilized CPU resource without affecting the performance of both collocated jobs. Based on this insight, an experimental system with collocation-aware job scheduler and resource manager is proposed. With our experiment workload pool of mixed CPU and GPU jobs, the system demonstrates 15% gain in throughput and 10% gain in both CPU and GPU utilization.

The MapReduce programming model, along with its open-source implementation Hadoop has provided a cost effective solution for many data-intensive applications. Hadoop stores data distributively and exploits data locality by assigning tasks to where data is stored. In many cases, however, accessing remote data (rack-local and off-rack) is inevitable. In this paper we are evaluating the possibility of improving the remote data accessing performance by streaming data from multiple available replicas. The proposed design consists of a circular buffer, a slice reader and a enhanced Data Node. Such system is capable of adapting to both the static performance variance caused by network topology as well as dynamic variance caused by congestion. Extensive experiments show that mutil-source streaming can significantly improve the throughput of remote data access and accelerate the related map tasks by 10%-20%. In some imbalanced environment, the proposed system can even achieve as much as 4x speedup.

Protein-DNA docking is a very challenging problem in bioinformatics and has important implications in a number of applications (e.g. rational drug design). This paper presents a computational approach to improve the prediction accuracy of protein-DNA docking. One of the major difficulties in protein-DNA docking is the high cost of sampling the conformational space. To address this problem, we develop a graphics processing unit (GPU)-based approach to accelerate the sampling process, and thereby improving the quality of conformational space sampling. The effectiveness of the our approach is validated against a non-redundant set of 75 protein-DNA complexes, and the results demonstrate improved performance in finding near-native protein-DNA complex structures. To the best of our knowledge, this is the first ad hoc effort of applying GPU or GPU clusters to the protein-DNA docking problem.