Search

Recent advances in generative diffusion models have enabled text-controlled synthesis of realistic and diverse images with impressive quality. Despite these remarkable advances, the application of text-to-image generative models in computer vision for standard visual recognition tasks remains limited. The current de facto approach for these tasks is to design model architectures and loss functions that are tailored to the task at hand. In this paper, we develop a unified language interface for computer vision tasks that abstracts away task-specific design choices and enables task execution by following natural language instructions. Our approach involves casting multiple computer vision tasks as text-to-image generation problems. Here, the text represents an instruction describing the task, and the resulting image is a visually-encoded task output. To train our model, we pool commonly-used computer vision datasets covering a range of tasks, including segmentation, object detection, depth estimation, and classification. We then use a large language model to paraphrase prompt templates that convey the specific tasks to be conducted on each image, and through this process, we create a multi-modal and multi-task training dataset comprising input and output images along with annotated instructions. Following the InstructPix2Pix architecture, we apply instruction-tuning to a text-to-image diffusion model using our constructed dataset, steering its functionality from a generative model to an instruction-guided multi-task vision learner. Experiments demonstrate that our model, dubbed InstructCV, performs competitively compared to other generalist and task-specific vision models. Moreover, it exhibits compelling generalization capabilities to unseen data, categories, and user instructions., Comment: ICLR 2024; Code is available at https://github.com/AlaaLab/InstructCV

Obtaining annotations for complex computer vision tasks such as object detection is an expensive and time-intense endeavor involving a large number of human workers or expert opinions. Reducing the amount of annotations required while maintaining algorithm performance is, therefore, desirable for machine learning practitioners and has been successfully achieved by active learning algorithms. However, it is not merely the amount of annotations which influences model performance but also the annotation quality. In practice, the oracles that are queried for new annotations frequently contain significant amounts of noise. Therefore, cleansing procedures are oftentimes necessary to review and correct given labels. This process is subject to the same budget as the initial annotation itself since it requires human workers or even domain experts. Here, we propose a composite active learning framework including a label review module for deep object detection. We show that utilizing part of the annotation budget to correct the noisy annotations partially in the active dataset leads to early improvements in model performance, especially when coupled with uncertainty-based query strategies. The precision of the label error proposals has a significant influence on the measured effect of the label review. In our experiments we achieve improvements of up to 4.5 mAP points of object detection performance by incorporating label reviews at equal annotation budget.

We capture optimal decay for the Mullins-Sekerka evolution, a nonlocal, parabolic free boundary problem from materials science. Our main result establishes convergence of BV solutions to the planar profile in the physically relevant case of ambient space dimension three. Far from assuming small or well-prepared initial data, we allow for initial interfaces that do not have graph structure and are not connected, hence explicitly including the regime of Ostwald ripening. In terms only of initially finite (not small) excess mass and excess surface energy, we establish that the surface becomes a Lipschitz graph within a fixed timescale (quantitatively estimated) and remains trapped within this setting. To obtain the graph structure, we leverage regularity results from geometric measure theory. At the same time, we extend a duality method previously employed for one-dimensional PDE problems to higher dimensional, nonlocal geometric evolutions. Optimal algebraic decay rates of excess energy, dissipation, and graph height are obtained., Comment: 48 pages - comments welcome, streamlined version, corrected insignificant sign mistakes, extended appendix

The evaluation of clustering results is difficult, highly dependent on the evaluated data set and the perspective of the beholder. There are many different clustering quality measures, which try to provide a general measure to validate clustering results. A very popular measure is the Silhouette. We discuss the efficient medoid-based variant of the Silhouette, perform a theoretical analysis of its properties, provide two fast versions for the direct optimization, and discuss the use to choose the optimal number of clusters. We combine ideas from the original Silhouette with the well-known PAM algorithm and its latest improvements FasterPAM. One of the versions guarantees equal results to the original variant and provides a run speedup of $O(k^2)$. In experiments on real data with 30000 samples and $k$=100, we observed a 10464$\times$ speedup compared to the original PAMMEDSIL algorithm. Additionally, we provide a variant to choose the optimal number of clusters directly., Comment: arXiv admin note: substantial text overlap with arXiv:2209.12553

Galaxy clusters are expected to be dark matter (DM) reservoirs and storage rooms for the cosmic-ray protons (CRp) that accumulate along the cluster's formation history. Accordingly, they are excellent targets to search for signals of DM annihilation and decay at gamma-ray energies and are predicted to be sources of large-scale gamma-ray emission due to hadronic interactions in the intracluster medium. We estimate the sensitivity of the Cherenkov Telescope Array (CTA) to detect diffuse gamma-ray emission from the Perseus galaxy cluster. We perform a detailed spatial and spectral modelling of the expected signal for the DM and the CRp components. For each, we compute the expected CTA sensitivity. The observing strategy of Perseus is also discussed. In the absence of a diffuse signal (non-detection), CTA should constrain the CRp to thermal energy ratio within the radius $R_{500}$ down to about $X_{500}<3\times 10^{-3}$, for a spatial CRp distribution that follows the thermal gas and a CRp spectral index $\alpha_{\rm CRp}=2.3$. Under the optimistic assumption of a pure hadronic origin of the Perseus radio mini-halo and depending on the assumed magnetic field profile, CTA should measure $\alpha_{\rm CRp}$ down to about $\Delta\alpha_{\rm CRp}\simeq 0.1$ and the CRp spatial distribution with 10% precision. Regarding DM, CTA should improve the current ground-based gamma-ray DM limits from clusters observations on the velocity-averaged annihilation cross-section by a factor of up to $\sim 5$, depending on the modelling of DM halo substructure. In the case of decay of DM particles, CTA will explore a new region of the parameter space, reaching models with $\tau_{\chi}>10^{27}$s for DM masses above 1 TeV. These constraints will provide unprecedented sensitivity to the physics of both CRp acceleration and transport at cluster scale and to TeV DM particle models, especially in the decay scenario., Comment: 93 pages (including author list, appendix and references), 143 figures. Submitted to JCAP

Partitioning Around Medoids (PAM, k-Medoids) is a popular clustering technique to use with arbitrary distance functions or similarities, where each cluster is represented by its most central object, called the medoid or the discrete median. In operations research, this family of problems is also known as facility location problem (FLP). FastPAM recently introduced a speedup for large k to make it applicable for larger problems, but the method still has a runtime quadratic in N. In this chapter, we discuss a sparse and asymmetric variant of this problem, to be used for example on graph data such as road networks. By exploiting sparsity, we can avoid the quadratic runtime and memory requirements, and make this method scalable to even larger problems, as long as we are able to build a small enough graph of sufficient connectivity to perform local optimization. Furthermore, we consider asymmetric cases, where the set of medoids is not identical to the set of points to be covered (or in the interpretation of facility location, where the possible facility locations are not identical to the consumer locations). Because of sparsity, it may be impossible to cover all points with just k medoids for too small k, which would render the problem unsolvable, and this breaks common heuristics for finding a good starting condition. We, hence, consider determining k as a part of the optimization problem and propose to first construct a greedy initial solution with a larger k, then to optimize the problem by alternating between PAM-style "swap" operations where the result is improved by replacing medoids with better alternatives and "remove" operations to reduce the number of k until neither allows further improving the result quality. We demonstrate the usefulness of this method on a problem from electrical engineering, with the input graph derived from cartographic data.

We present a simulation capability for micro-scale light-emitting diodes (uLEDs) that achieves comparable accuracy to CPU-based finite-difference time-domain simulation but is more than 10^7 times faster. Our approach is based on the Fourier modal method (FMM) -- which, as we demonstrate, is well suited to modeling thousands of incoherent sources -- with extensions that allow rapid convergence for uLED structures that are challenging to model with standard approaches. The speed of our method makes the inverse design of uLEDs tractable, which we demonstrate by designing a metasurface-enhanced uLED that doubles the light extraction efficiency of an unoptimized device., Comment: 15 pages, 10 figures

One aim of Process Mining (PM) is the discovery of process models from event logs of information systems. PM has been successfully applied to process-oriented enterprise systems but is less suited for communication- and document-oriented Enterprise Collaboration Systems (ECS). ECS event logs are very fine-granular and PM applied to their logs results in spaghetti models. A common solution for this is event abstraction, i.e., converting low-level logs into more abstract high-level logs before running discovery algorithms. ECS logs come with special characteristics that have so far not been fully addressed by existing event abstraction approaches. We aim to close this gap with a tailored ECS event abstraction (ECSEA) approach that trains a model by comparing recorded actual user activities (high-level traces) with the system-generated low-level traces (extracted from the ECS). The model allows us to automatically convert future low-level traces into an abstracted high-level log that can be used for PM. Our evaluation shows that the algorithm produces accurate results. ECSEA is a preprocessing method that is essential for the interpretation of collaborative work activity in ECS, which we call Social Process Mining., Comment: 8 pages, 1 figure, 3 tables

We consider an open quantum system described by the GKLS equation and we are interested in the onset of decoherence. We are in particulary interested in situations where only some degrees of freedom of the system are coupled to the environment, and we want to understand if, and how fast, the noise travels through the system and eventually affects all degrees of freedom. We find that this can be understood in terms of the H\"ormander condition, a condition on the commutators of the Hamiltonian vectorfields of the Lindblad operators and the internal Hamiltonian, which is a condition for hypoellipticity known from the theory of PDE's. We show that for Gaussian quantum channels this condition leads to a delay in the onset of decoherence and can as well be used to detect decoherence free subsystems., Comment: 9 pages