Automated machine learning(AutoML) has achieved remarkable success in automating the non-trivial process of designing machine learning *** the focal areas of AutoML,neural architecture search(NAS) stands out,aiming to...
详细信息
Automated machine learning(AutoML) has achieved remarkable success in automating the non-trivial process of designing machine learning *** the focal areas of AutoML,neural architecture search(NAS) stands out,aiming to systematically explore the complex architecture space to discover the optimal neural architecture configurations without intensive manual *** has demonstrated its capability of dramatic performance improvement across a large number of real-world *** core components in NAS methodologies normally include(ⅰ) defining the appropriate search space,(ⅱ)designing the right search strategy and(ⅲ) developing the effective evaluation *** early NAS endeavors are characterized via groundbreaking architecture designs,the imposed exorbitant computational demands prompt a shift towards more efficient paradigms such as weight sharing and evaluation estimation,***,the introduction of specialized benchmarks has paved the way for standardized comparisons of NAS ***,the adaptability of NAS is evidenced by its capability of extending to diverse datasets,including graphs,tabular data and videos,etc.,each of which requires a tailored *** paper delves into the multifaceted aspects of NAS,elaborating on its recent advances,applications,tools,benchmarks and prospective research directions.
Video question answering(VideoQA) is a challenging yet important task that requires a joint understanding of low-level video content and high-level textual semantics. Despite the promising progress of existing efforts...
详细信息
Video question answering(VideoQA) is a challenging yet important task that requires a joint understanding of low-level video content and high-level textual semantics. Despite the promising progress of existing efforts, recent studies revealed that current VideoQA models mostly tend to over-rely on the superficial correlations rooted in the dataset bias while overlooking the key video content, thus leading to unreliable results. Effectively understanding and modeling the temporal and semantic characteristics of a given video for robust VideoQA is crucial but, to our knowledge, has not been well investigated. To fill the research gap, we propose a robust VideoQA framework that can effectively model the cross-modality fusion and enforce the model to focus on the temporal and global content of videos when making a QA decision instead of exploiting the shortcuts in datasets. Specifically, we design a self-supervised contrastive learning objective to contrast the positive and negative pairs of multimodal input, where the fused representation of the original multimodal input is enforced to be closer to that of the intervened input based on video perturbation. We expect the fused representation to focus more on the global context of videos rather than some static keyframes. Moreover, we introduce an effective temporal order regularization to enforce the inherent sequential structure of videos for video representation. We also design a Kullback-Leibler divergence-based perturbation invariance regularization of the predicted answer distribution to improve the robustness of the model against temporal content perturbation of videos. Our method is model-agnostic and can be easily compatible with various VideoQA backbones. Extensive experimental results and analyses on several public datasets show the advantage of our method over the state-of-the-art methods in terms of both accuracy and robustness.
Data race is one of the most important concurrent anomalies in multi-threaded *** con-straint-based techniques are leveraged into race detection,which is able to find all the races that can be found by any oth-er soun...
详细信息
Data race is one of the most important concurrent anomalies in multi-threaded *** con-straint-based techniques are leveraged into race detection,which is able to find all the races that can be found by any oth-er sound race ***,this constraint-based approach has serious limitations on helping programmers analyze and understand data ***,it may report a large number of false positives due to the unrecognized dataflow propa-gation of the ***,it recommends a wide range of thread context switches to schedule the reported race(in-cluding the false one)whenever this race is exposed during the constraint-solving *** ad hoc recommendation imposes too many context switches,which complicates the data race *** address these two limitations in the state-of-the-art constraint-based race detection,this paper proposes DFTracker,an improved constraint-based race detec-tor to recommend each data race with minimal thread context ***,we reduce the false positives by ana-lyzing and tracking the dataflow in the *** this means,DFTracker thus reduces the unnecessary analysis of false race *** further propose a novel algorithm to recommend an effective race schedule with minimal thread con-text switches for each data *** experimental results on the real applications demonstrate that 1)without removing any true data race,DFTracker effectively prunes false positives by 68%in comparison with the state-of-the-art constraint-based race detector;2)DFTracker recommends as low as 2.6-8.3(4.7 on average)thread context switches per data race in the real world,which is 81.6%fewer context switches per data race than the state-of-the-art constraint based race ***,DFTracker can be used as an effective tool to understand the data race for programmers.
Machine learning has been massively utilized to construct data-driven solutions for predicting the lifetime of rechargeable batteries in recent years, which project the physical measurements obtained during the early ...
详细信息
Machine learning has been massively utilized to construct data-driven solutions for predicting the lifetime of rechargeable batteries in recent years, which project the physical measurements obtained during the early charging/discharging cycles to the remaining useful lifetime. While most existing techniques train the prediction model through minimizing the prediction error only, the errors associated with the physical measurements can also induce negative impact to the prediction accuracy. Although total-least-squares(TLS) regression has been applied to address this issue, it relies on the unrealistic assumption that the distributions of measurement errors on all input variables are equivalent, and cannot appropriately capture the practical characteristics of battery degradation. In order to tackle this challenge, this work intends to model the variations along different input dimensions, thereby improving the accuracy and robustness of battery lifetime prediction. In specific, we propose an innovative EM-TLS framework that enhances the TLS-based prediction to accommodate dimension-variate errors, while simultaneously investigating the distributions of them using expectation-maximization(EM). Experiments have been conducted to validate the proposed method based on the data of commercial Lithium-Ion batteries, where it reduces the prediction error by up to 29.9 % compared with conventional TLS. This demonstrates the immense potential of the proposed method for advancing the R&D of rechargeable batteries.
Graph processing has been widely used in many scenarios,from scientific computing to artificial *** processing exhibits irregular computational parallelism and random memory accesses,unlike traditional ***,running gra...
详细信息
Graph processing has been widely used in many scenarios,from scientific computing to artificial *** processing exhibits irregular computational parallelism and random memory accesses,unlike traditional ***,running graph processing workloads on conventional architectures(e.g.,CPUs and GPUs)often shows a significantly low compute-memory ratio with few performance benefits,which can be,in many cases,even slower than a specialized single-thread graph *** domain-specific hardware designs are essential for graph processing,it is still challenging to transform the hardware capability to performance boost without coupled software *** article presents a graph processing ecosystem from hardware to *** start by introducing a series of hardware accelerators as the foundation of this ***,the codesigned parallel graph systems and their distributed techniques are presented to support graph ***,we introduce our efforts on novel graph applications and hardware *** results show that various graph applications can be efficiently accelerated in this graph processing ecosystem.
Temporal knowledge graph(TKG) reasoning, has seen widespread use for modeling real-world events, particularly in extrapolation settings. Nevertheless, most previous studies are embedded models, which require both enti...
详细信息
Temporal knowledge graph(TKG) reasoning, has seen widespread use for modeling real-world events, particularly in extrapolation settings. Nevertheless, most previous studies are embedded models, which require both entity and relation embedding to make predictions, ignoring the semantic correlations among different entities and relations within the same timestamp. This can lead to random and nonsensical predictions when unseen entities or relations occur. Furthermore, many existing models exhibit limitations in handling highly correlated historical facts with extensive temporal depth. They often either overlook such facts or overly accentuate the relationships between recurring past occurrences and their current counterparts. Due to the dynamic nature of TKG, effectively capturing the evolving semantics between different timestamps can be *** address these shortcomings, we propose the recurrent semantic evidenceaware graph neural network(RE-SEGNN), a novel graph neural network that can learn the semantics of entities and relations simultaneously. For the former challenge, our model can predict a possible answer to missing quadruples based on semantics when facing unseen entities or relations. For the latter problem, based on an obvious established force, both the recency and frequency of semantic history tend to confer a higher reference value for the current. We use the Hawkes process to compute the semantic trend, which allows the semantics of recent facts to gain more attention than those of distant facts. Experimental results show that RE-SEGNN outperforms all SOTA models in entity prediction on 6 widely used datasets, and 5 datasets in relation prediction. Furthermore, the case study shows how our model can deal with unseen entities and relations.
This paper presents ScenePalette,a modeling tool that allows users to“draw”3D scenes interactively by placing objects on a canvas based on their contextual *** is inspired by an important intuition which was often i...
详细信息
This paper presents ScenePalette,a modeling tool that allows users to“draw”3D scenes interactively by placing objects on a canvas based on their contextual *** is inspired by an important intuition which was often ignored in previous work:a real-world 3D scene consists of the contextually reasonable organization of objects,*** typically place one double bed with several subordinate objects into a bedroom instead of different shapes of ***,abstracts 3D repositories as multiplex networks and accordingly encodes implicit relations between or among ***,basic statistics such as co-occurrence,in combination with advanced relations,are used to tackle object relationships of different *** experiments demonstrate that the latent space of ScenePalette has rich contexts that are essential for contextual representation and exploration.
The problem of imbalanced data classification learning has received much *** classification algorithms are susceptible to data skew to favor majority samples and ignore minority *** weighted minority oversampling tech...
详细信息
The problem of imbalanced data classification learning has received much *** classification algorithms are susceptible to data skew to favor majority samples and ignore minority *** weighted minority oversampling technique(MWMOTE)is an effective approach to solve this problem,however,it may suffer from the shortcomings of inadequate noise filtering and synthesizing the same samples as the original minority *** this end,we propose an improved MWMOTE method named joint sample position based noise filtering and mean shift clustering(SPMSC)to solve these ***,in order to effectively eliminate the effect of noisy samples,SPMsC uses a new noise filtering mechanism to determine whether a minority sample is noisy or not based on its position and distribution relative to the majority *** that MWMOTE may generate duplicate samples,we then employ the mean shift algorithm to cluster minority samples to reduce synthetic replicate ***,data cleaning is performed on the processed data to further eliminate class *** on extensive benchmark datasets demonstrate the effectiveness of SPMsC compared with other sampling methods.
Sufficient and high-quality data are a prerequisite for building complex machine learning systems, especially those with a large number of parameters (e.g., ChatGPT). However, it is typically challenging or even impos...
详细信息
Sufficient and high-quality data are a prerequisite for building complex machine learning systems, especially those with a large number of parameters (e.g., ChatGPT). However, it is typically challenging or even impossible to acquire a sufficient amount of real data to train such systems. For example, auto-driving systems may need to learn from various accidental events in order to be reliable in driving, while collecting such real data is difficult or ethically infeasible.
Hybrid memory systems composed of dynamic random access memory(DRAM)and Non-volatile memory(NVM)often exploit page migration technologies to fully take the advantages of different memory *** previous proposals usually...
详细信息
Hybrid memory systems composed of dynamic random access memory(DRAM)and Non-volatile memory(NVM)often exploit page migration technologies to fully take the advantages of different memory *** previous proposals usually migrate data at a granularity of 4 KB pages,and thus waste memory bandwidth and DRAM *** this paper,we propose Mocha,a non-hierarchical architecture that organizes DRAM and NVM in a flat address space physically,but manages them in a cache/memory *** the commercial NVM device-Intel Optane DC Persistent Memory Modules(DCPMM)actually access the physical media at a granularity of 256 bytes(an Optane block),we manage the DRAM cache at the 256-byte size to adapt to this feature of *** design not only enables fine-grained data migration and management for the DRAM cache,but also avoids write amplification for Intel Optane *** also create an Indirect Address Cache(IAC)in Hybrid Memory Controller(HMC)and propose a reverse address mapping table in the DRAM to speed up address translation and cache ***,we exploit a utility-based caching mechanism to filter cold blocks in the NVM,and further improve the efficiency of the DRAM *** implement Mocha in an architectural *** results show that Mocha can improve application performance by 8.2%on average(up to 24.6%),reduce 6.9%energy consumption and 25.9%data migration traffic on average,compared with a typical hybrid memory architecture-HSCC.
暂无评论