The generation of voluminous scientific data poses significant challenges for efficient storage, transfer, and analysis. Recently, error-bounded lossy compression methods emerged due to their ability to achieve high compression ratios while controlling data distortion. However, they often overlook the inherent spatial and temporal correlations within scientific data, thus missing opportunities for higher compression. In this paper we propose GRAPHCOMP, a novel graph-based method for error-bounded lossy compression of scientific data. We perform irregular segmentation of the original grid data and generate a graph representation that preserves the spatial and temporal correlations. Inspired by Graph Neural Networks (GNNs), we then propose a temporal graph autoencoder to learn latent representations that significantly reduce the size of the graph, effectively compressing the original data. Decompression reverses the process and utilizes the learnt graph model together with the latent representation to reconstruct an approximation of the original data. The decompressed data are guaranteed to satisfy a user-defined point-wise error bound. We compare our method against the state-of-the-art error-bounded lossy methods (i.e., HPEZ, SZ3.1, SPERR, and ZFP) on large-scale real and synthetic data. GRAPHCOMP consistently achieves the highest compression ratio across most datasets, outperforming the second-best method by margins ranging from 22% to 50%.

Time series similarity search (TSSS) is a fundamental task across various applications, including classification, motif discovery, and anomaly detection. However, existing iSAX-based index methods, while known for their efficiency, often rely on hand-crafted techniques (e.g., PAA and SAX) for z-normalized time series data. However, these techniques do not fully exploit the full representation space and pose challenges to indexing. In this paper, we propose a learned index approach for TSSS. Specifically, we introduce SAXnet, a novel two-stage neural network that generates the learned SAX representation (leSAX representation) for both z-normalized and non-z-normalized time series data. The benefits of SAXnet are threefold: ① full exploitation of latent space, ② preservation of time series shapes and global information for indexing, and ③ elimination of the need for hand-crafted techniques. We then propose leSaxindex, a novel learned SAX representation index, which consists of a leSAX tree and a learned index. The distribution of the leSAX representations in the leSAX tree is adjusted to achieve a near-uniform distribution for index efficiency. Furthermore, we propose a learned index structure that works alongside the leSAX tree, applied recursively in case of large index leaf nodes. We have conducted comprehensive experiments on exact similarity search using our SAXnet and leSAX index on both real and synthetic time series datasets. The results demonstrate that our leSAX method outperforms state-of-the-art methods in efficiency, achieving performance improvements ranging from 3.6× to 17×.

Transformer-based models have facilitated numerous applications with superior performance. A key challenge in transformers is the quadratic dependency of its training time complexity on the length of the input sequence. A recent popular solution is using random feature attention (RFA) to approximate the costly vanilla attention mechanism. However, RFA relies on only a single, fixed projection for approximation, which does not capture the input distribution and can lead to low efficiency and accuracy, especially on time series data. In this paper, we propose DARKER, an efficient transformer with a novelDAta-dRivenKERnel-based attention mechanism. To precisely present the technical details, this paper discusses them with a fundamental time series task, namely, time series classification (tsc). First, the main novelty of DARKER lies in approximating the softmax kernel by learning multiple machine learning models with trainable weights as multiple projections offline, moving beyond the limitation of a fixed projection. Second, we propose a projection index (called pIndex) to efficiently search the most suitable projection for the input for training transformer. As a result, the overall time complexity of DARKER is linear with the input length. Third, we propose an indexing technique for efficiently computing the inputs required for transformer training. Finally, we evaluate our method on 14 real-world and 2 synthetic time series datasets. The experiments show that DARKER is 3×-4× faster than vanilla transformer and 1.5×-3× faster than other SOTAs for long sequences. In addition, the accuracy of DARKER is comparable to or higher than that of all compared transformers.

Time series classification (TSC) has been one of the most fundamental problems of time series data. Time series shapelets (or simply, shapelets) are discriminative subsequences that have been recently found both effective and interpretable for solving TSC. However, shapelet discovery is known to be computationally costly. Meanwhile, matrix profile has been recently proposed for efficient motif discovery and anomaly detection. Our preliminary experiment shows that a direct adoption of the matrix profile on TSC does not bring superior classification accuracy. We have identified two main issues of such an adoption: 1) discords as “shapelets”, and 2) lack of shapelet diversity. In response to these issues, we propose instance profile for shapelets, called IPS, for shapelet discovery for TSC. The main challenge is to utilize the instance profile (IP) to capture the characteristics of shapelets in a robust manner and then to discover high-quality shapelets efficiently. First, we use our IP to generate abundant shapelet candidates. We next efficiently prune candidates that do not align with the definition of shapelets using a novel distribution-aware bloom filter (DABF). Three utility functions are proposed to measure the shapelet candidates and DABF is used to efficiently compute the functions. We have conducted comprehensive experiments on IPS with 12 competitive state-of-the-art methods using UCR Archive datasets. The efficiency is on average 25 times faster than that of BSPCOVER (the current state-of-the-art method). The accuracy of IPS is comparable to or higher than that of existing work. Furthermore, we select one case study to illustrate the interpretability of the shapelets.

Sketching is a common approach to visually query time series data. However, a recent study reported that sketching a pattern for querying is "often ineffective on its own" in practice due to lack of "representative objects" to facilitate bottom-up search. In this demonstration, we present a novel data-driven sketch-based visual query interface (VQI) construction system called SENSOR to alleviate this challenge. Given a time series dataset, SENSOR automatically constructs its VQI by populating different components from the underlying data. Specifically, it discovers and exposes a set of representative objects in the form of VST-aware shapelets to facilitate query formulation. Such data-driven construction has several potential benefits such as empowering efficient top-down and bottom-up search and portability of the interface across different application domains and sources.

Time-series shapelets are discriminative subsequences, recently found effective for time series classification (tsc). It is evident that the quality of shapelets is crucial to the accuracy oftsc. However, major research has focused on building accurate models from some shapelet candidates. To determine such candidates, existing studies are surprisingly simple, e.g., enumerating subsequences of some fixed lengths, or randomly selecting some subsequences as shapelet candidates. The major bulk of computation is then on building the model from the candidates. In this paper, we propose a novelefficient shapelet discoverymethod, calledbspcover, to discover a set of high-quality shapelet candidates for model building. Specifically,bspcovergenerates abundant candidates via Symbolic Aggregate approXimation with sliding window, then prunes identical and highly similar candidates viaBloom filters, andsimilarity matching, respectively. We next propose a$p$p-Cover algorithmto efficiently determine discriminative shapelet candidates that maximally represent each time-series class. Finally, any existing shapelet learning method can be adopted to build a classification model. We have conducted extensive experiments with well-known time-series datasets and representative state-of-the-art methods. Results show thatbspcoverspeeds up the state-of-the-art methods by more than 70 times, and the accuracy is often comparable to or higher than existing works.

Time-series shapelets are discriminative subsequences, recently found effective for time series classification (TSC). It is evident that the quality of shapelets is crucial to the accuracy of TSC. However, major research has focused on building accurate models from some shapelet candidates. To determine such candidates, existing studies are surprisingly simple, e.g., enumerating subsequences of some fixed lengths, or randomly selecting some subsequences as shapelet candidates. The major bulk of computation is then on building the model from the candidates. In this paper, we propose a novel efficient shapelet discovery method, called BSPCOVER, to discover a set of high-quality shapelet candidates for model building. We have conducted extensive experiments with well-known UCR time-series datasets and representative state-of-the-art methods. Results show that BSPCOVER speeds up the state-of-the-art methods by more than 70 times, and the accuracy is often comparable to or higher than existing works.

Time series classification (TSC) has attracted considerable attention from both academia and industry. TSC methods that are based on shapelets (intuitively, small highly-discriminative subsequences have been found effective and are particularly known for their interpretability, as shapelets themselves are subsequences. A recent work has significantly improved the efficiency of shapelet discovery. For instance, the shapelets of more than 65% of the datasets in the UCR Archive (containing data from different application domains) can be computed within an hour, whereas those of 12 datasets can be computed within a minute. Such efficiency has made it possible for demo attendees to interact with shapelet discovery and explore high-quality shapelets. In this demo, we present Visualet -- a tool for visualizing shapelets, and exploring effective and interpretable ones.