Search

Data dependency-based query optimization techniques can considerably improve database system performance: we apply three such optimization techniques to five database management systems (DBMSs) and observe throughput improvements between 5 % and 33 %. We address two key challenges to achieve these results: (i) efficiently identifying and extracting relevant dependencies from the data, and (ii) making use of the dependencies through SQL rewrites or as transformation rules in the optimizer. First, the schema does not provide all relevant dependencies. We present a workload-driven dependency discovery approach to find additional dependencies within milliseconds. Second, the throughput improvement of a state-of-the-art DBMS is 13 % using only SQL rewrites, but 20 % when we integrate dependency-based optimization into the optimizer and execution engine, e. g., by employing dependency propagation and subquery handling. Using all relevant dependencies, the runtime of four standard benchmarks improves by up to 10 % compared to using only primary and foreign keys, and up to 22 % compared to not using dependencies. The dependency discovery overhead amortizes after a single workload execution.

A typical process of learning to play a piece on a piano consists of a progression through a series of practice units that focus on individual dimensions of the skill, the so-called practice modes. Practice modes in learning to play music comprise a particularly large set of possibilities, such as hand coordination, posture, articulation, ability to read a music score, correct timing or pitch, etc. Self-guided practice is known to be suboptimal, and a model that schedules optimal practice to maximize a learner's progress still does not exist. Because we each learn differently and there are many choices for possible piano practice tasks and methods, the set of practice modes should be dynamically adapted to the human learner, a process typically guided by a teacher. However, having a human teacher guide individual practice is not always feasible since it is time-consuming, expensive, and often unavailable. In this work, we present a modeling framework to guide the human learner through the learning process by choosing the practice modes generated by a policy model. To this end, we present a computational architecture building on a Gaussian process that incorporates 1) the learner state, 2) a policy that selects a suitable practice mode, 3) performance evaluation, and 4) expert knowledge. The proposed policy model is trained to approximate the expert-learner interaction during a practice session. In our future work, we will test different Bayesian optimization techniques, e.g., different acquisition functions, and evaluate their effect on the learning progress.

Deep learning models for speech rely on large datasets, presenting computational challenges. Yet, performance hinges on training data size. Dataset Distillation (DD) aims to learn a smaller dataset without much performance degradation when training with it. DD has been investigated in computer vision but not yet in speech. This paper presents the first approach for DD to speech targeting Speech Emotion Recognition on IEMOCAP. We employ Generative Adversarial Networks (GANs) not to mimic real data but to distil key discriminative information of IEMOCAP that is useful for downstream training. The GAN then replaces the original dataset and can sample custom synthetic dataset sizes. It performs comparably when following the original class imbalance but improves performance by 0.3% absolute UAR with balanced classes. It also reduces dataset storage and accelerates downstream training by 95% in both cases and reduces speaker information which could help for a privacy application., Comment: Accepted at Interspeech 2024

This paper introduces TVB C++, a streamlined and fast C++ Back-End for The Virtual Brain (TVB), a renowned platform and a benchmark tool for full-brain simulation. TVB C++ is engineered with speed as a primary focus while retaining the flexibility and ease of use characteristic of the original TVB platform. Positioned as a complementary tool, TVB serves as a prototyping platform, whereas TVB C++ becomes indispensable when performance is paramount, particularly for large-scale simulations and leveraging advanced computation facilities like supercomputers. Developed as a TVB-compatible Back-End, TVB C++ seamlessly integrates with the original TVB implementation, facilitating effortless usage. Users can easily configure TVB C++ to execute the same code as in TVB but with enhanced performance and parallelism capabilities.

Given a finite graph $G$, we define the Ceresa period $\alpha(G)$ as a tool for studying algebraic triviality of the tropical Ceresa cycle introduced by Zharkov. We show that $\alpha(G) = 0$ if and only if $G$ is of hyperelliptic type; then a theorem of Corey implies that having $\alpha(G) = 0$ is a minor-closed condition with forbidden minors $K_4$ and $L_3$.

Intelligent Tutoring Systems (ITSs) often contain an automated feedback component, which provides a predefined feedback message to students when they detect a predefined error. To such a feedback component, we often resort to template-based approaches. These approaches require significant effort from human experts to detect a limited number of possible student errors and provide corresponding feedback. This limitation is exemplified in open-ended math questions, where there can be a large number of different incorrect errors. In our work, we examine the capabilities of large language models (LLMs) to generate feedback for open-ended math questions, similar to that of an established ITS that uses a template-based approach. We fine-tune both open-source and proprietary LLMs on real student responses and corresponding ITS-provided feedback. We measure the quality of the generated feedback using text similarity metrics. We find that open-source and proprietary models both show promise in replicating the feedback they see during training, but do not generalize well to previously unseen student errors. These results suggest that despite being able to learn the formatting of feedback, LLMs are not able to fully understand mathematical errors made by students., Comment: Educational Data Mining 2024

Diffusion generative models have recently become a powerful technique for creating and modifying high-quality, coherent video content. This survey provides a comprehensive overview of the critical components of diffusion models for video generation, including their applications, architectural design, and temporal dynamics modeling. The paper begins by discussing the core principles and mathematical formulations, then explores various architectural choices and methods for maintaining temporal consistency. A taxonomy of applications is presented, categorizing models based on input modalities such as text prompts, images, videos, and audio signals. Advancements in text-to-video generation are discussed to illustrate the state-of-the-art capabilities and limitations of current approaches. Additionally, the survey summarizes recent developments in training and evaluation practices, including the use of diverse video and image datasets and the adoption of various evaluation metrics to assess model performance. The survey concludes with an examination of ongoing challenges, such as generating longer videos and managing computational costs, and offers insights into potential future directions for the field. By consolidating the latest research and developments, this survey aims to serve as a valuable resource for researchers and practitioners working with video diffusion models. Website: https://github.com/ndrwmlnk/Awesome-Video-Diffusion-Models, Comment: https://github.com/ndrwmlnk/Awesome-Video-Diffusion-Models

Focusing is a known technique for reducing the number of proofs while preserving derivability. Skolemisation is another technique designed to improve proof search, which reduces the number of back-tracking steps by representing dependencies on the term level and instantiate witness terms during unification at the axioms or fail with an occurs-check otherwise. Skolemisation for classical logic is well understood, but a practical skolemisation procedure for focused intuitionistic linear logic has been elusive so far. In this paper we present a focused variant of first-order intuitionistic linear logic together with a sound and complete skolemisation procedure., Comment: to appear at IJCAR 2024

The lane graph is a key component for building high-definition (HD) maps and crucial for downstream tasks such as autonomous driving or navigation planning. Previously, He et al. (2022) explored the extraction of the lane-level graph from aerial imagery utilizing a segmentation based approach. However, segmentation networks struggle to achieve perfect segmentation masks resulting in inaccurate lane graph extraction. We explore additional enhancements to refine this segmentation-based approach and extend it with a diffusion probabilistic model (DPM) component. This combination further improves the GEO F1 and TOPO F1 scores, which are crucial indicators of the quality of a lane graph, in the undirected graph in non-intersection areas. We conduct experiments on a publicly available dataset, demonstrating that our method outperforms the previous approach, particularly in enhancing the connectivity of such a graph, as measured by the TOPO F1 score. Moreover, we perform ablation studies on the individual components of our method to understand their contribution and evaluate their effectiveness.

We develop comparison results for Markov tree distributions extending ordering results from the literature on discrete time Markov processes and recently studied ordering results for conditionally independent factor models to tree structures. Based on fairly natural positive dependence conditions, our main contribution is a comparison result with respect to the supermodular order. Since this order is a pure dependence order, it has many applications in optimal transport, finance, and insurance. As an illustrative example, we consider hidden Markov models and study distributional robustness for functionals of the random walk under model uncertainty. Further, we show that, surprisingly, more general comparison results via the recently established rearrangement-based Schur order for conditional distributions, which implies an ordering of Chatterjee's rank correlation, do not carry over from star structures to trees. Several examples and a detailed discussion of the assumptions demonstrate the generality of our results and provide further insights into the behavior of multidimensional distributions.

To properly solve the coincidence problem ($\Omega_\mathrm{DM} \simeq 5\Omega_\mathrm{VM}$) in a model of asymmetric dark matter, one cannot simply relate the number densities of visible and dark matter without also relating their particle masses. Following previous work, we consider a framework where the dark matter is a confined state of a dark QCD gauge group whose confinement scale is dynamically related to the QCD confinement scale by a mechanism utilising infrared fixed points of the two gauge couplings. In this work we present a new, `zero-coupling infrared fixed point' approach, which allows a larger proportion of models in this framework to generically relate the masses of the visible and dark matter particles. Due to the heavy mass scale required for the new field content, we introduce supersymmetry to the theory. We consider how these models may be incorporated in a full theory of asymmetric dark matter, presenting some example leptogenesis-like models. We also discuss the phenomenology of these models; in particular, there are gravitational wave signals which, while weak, may be measurable at future mHz and $\mu$Hz detectors., Comment: 28 pages, 9 figures, 8 tables

A key challenge towards future quantum internet technology is connecting quantum processors at metropolitan scale. Here, we report on heralded entanglement between two independently operated quantum network nodes separated by 10km. The two nodes hosting diamond spin qubits are linked with a midpoint station via 25km of deployed optical fiber. We minimize the effects of fiber photon loss by quantum frequency conversion of the qubit-native photons to the telecom L-band and by embedding the link in an extensible phase-stabilized architecture enabling the use of the loss-resilient single-photon entangling protocol. By capitalizing on the full heralding capabilities of the network link in combination with real-time feedback logic on the long-lived qubits, we demonstrate the delivery of a predefined entangled state on the nodes irrespective of the heralding detection pattern. Addressing key scaling challenges and being compatible with different qubit systems, our architecture establishes a generic platform for exploring metropolitan-scale quantum networks., Comment: 10 pages, 4 figures, supplementary materials

Transformer-based language models spread FLOPs uniformly across input sequences. In this work we demonstrate that transformers can instead learn to dynamically allocate FLOPs (or compute) to specific positions in a sequence, optimising the allocation along the sequence for different layers across the model depth. Our method enforces a total compute budget by capping the number of tokens ($k$) that can participate in the self-attention and MLP computations at a given layer. The tokens to be processed are determined by the network using a top-$k$ routing mechanism. Since $k$ is defined a priori, this simple procedure uses a static computation graph with known tensor sizes, unlike other conditional computation techniques. Nevertheless, since the identities of the $k$ tokens are fluid, this method can expend FLOPs non-uniformly across the time and model depth dimensions. Thus, compute expenditure is entirely predictable in sum total, but dynamic and context-sensitive at the token-level. Not only do models trained in this way learn to dynamically allocate compute, they do so efficiently. These models match baseline performance for equivalent FLOPS and wall-clock times to train, but require a fraction of the FLOPs per forward pass, and can be upwards of 50\% faster to step during post-training sampling.

Radar sensors have a long tradition in advanced driver assistance systems (ADAS) and also play a major role in current concepts for autonomous vehicles. Their importance is reasoned by their high robustness against meteorological effects, such as rain, snow, or fog, and the radar's ability to measure relative radial velocity differences via the Doppler effect. The cause for these advantages, namely the large wavelength, is also one of the drawbacks of radar sensors. Compared to camera or lidar sensor, a lot more surfaces in a typical traffic scenario appear flat relative to the radar's emitted signal. This results in multi-path reflections or so called ghost detections in the radar signal. Ghost objects pose a major source for potential false positive detections in a vehicle's perception pipeline. Therefore, it is important to be able to segregate multi-path reflections from direct ones. In this article, we present a dataset with detailed manual annotations for different kinds of ghost detections. Moreover, two different approaches for identifying these kinds of objects are evaluated. We hope that our dataset encourages more researchers to engage in the fields of multi-path object suppression or exploitation.