搜索结果：dinosaurs

Slot-based object-centric learning represents an image as a set of latent slots with a decoder that combines them into an image or features. The decoder specifies how slots are combined into an output, but the slot set is typically fixed: the number of slots is chosen upfront and slots are only refined. This can lead to multiple slots competing for overlapping regions of the same entity rather than focusing on distinct regions. We introduce slot merging: a drop-in, lightweight operation on the slot set that merges overlapping slots during training. We quantify overlap with a Soft-IoU score between slot-attention maps and combine selected pairs via a barycentric update that preserves gradient flow. Merging follows a fixed policy, with the decision threshold inferred from overlap statistics, requiring no additional learnable modules. Integrated into the established feature-reconstruction pipeline of DINOSAUR, the proposed method improves object factorization and mask quality, surpassing other adaptive methods in object discovery and segmentation benchmarks.

Approximate Nearest Neighbour search indices form the backbone of real-world recommender systems, enabling real-time candidate retrieval over million-item catalogues. Typically, a single point estimate embedding is learnt for every user and every item. At serving time, the user embedding queries the index for relevant items. Since these representations are learnt from sparse interaction data, they are noisy and might fail to capture all the nuances that contribute to ``relevance'' -- ignoring the fundamental uncertainty that is inherent to them. The result is a retrieval pipeline that is systematically biased toward the small minority of popular head items with well-estimated embeddings, at the expense of the long-tail majority of niche, diverse, and serendipitous content. We propose DINOSAUR (Distributional Approximate Nearest Neighbour Search for Uncertainty-Aware Retrieval): a simple and infrastructure-compatible framework to incorporate embedding uncertainty into candidate generation. Rather than indexing point estimates, DINOSAUR samples $S_i$ embeddings per item and constructs an index on this augmented set. Analogously, at query time, a user embedding is sampled. This two-si

Seventeen representative samples of volcanic origin were collected from Ecuador (Pichincha Volcano), Iceland (Eyjafjallajökull Volcano), India (Deccan Traps), Hawaii, Kilimanjaro, Mt. Etna, Rwanda (Virunga Mountains), and Uganda (Virunga Mountains). Neutron activation analysis (NAA) was performed to determine the concentration of 33 chemical elements, including 21 trace elements, 20 heavy metals, and 9 rare earth elements: Al, As, Ba, Ca, Ce, Cl, Co, Cr, Cs, Dy, Eu, Fe, Hf, K, La, Lu, Mg, Mn, Na, Nd, Rb, Sb, Sc, Sm, Sr, Ta, Tb, Th, Ti, U, Yb, Zn, and Zr. Correlation analysis of the abundance of samples from different islands in the Hawaii archipelago (Kauai, Kilauea, Mauna Loa, and Haleakala) confirmed that the islands were likely formed by two different lava sources. Additionally, the upper limit of iridium was determined in 11 of these samples using Bayesian analysis, which does not support the hypothesis that volcanic activity caused the extinction of the dinosaurs. We also discuss how the abundance of thorium and uranium in lava from different geological formations and depths can contribute to building a better map of natural radioisotope occurrences on Earth, which is importan

Continual anomaly detection (CAD) addresses the need for industrial inspection systems to adapt to evolving production conditions, yet existing methods share three critical gaps: unrealistic evaluation, no systematic comparison, and no consideration of edge deployment constraints. We introduce a unified benchmark combining discrete-task evaluation on structural and logical anomalies, a novel continuous drift protocol, the first head-to-head comparison of all published CAD methods, and computational efficiency profiling on edge hardware. Our results reveal that existing CAD methods do not consistently outperform traditional approaches with simple experience replay. Thus motivated, we propose DINOSaur, a training-free method combining a frozen DINOv3 backbone with spatially-indexed coreset memory and neighborhood-restricted anomaly scoring. DINOSaur achieves zero forgetting by construction, outperforms all evaluated methods across all five protocols, and runs at sub-100\,ms inference on an NVIDIA Jetson Orin Nano, with on-device adaptation to new tasks in under 30 seconds.

Conversational agents, such as chatbots, have increasingly found their way into many dimensions of our lives, including entertainment and education. In this exploratory study we built a child-friendly chatbot, "Ask Me Anything" (AMA), and investigated children's attitudes and trust toward AI-driven conversational agents. To prompt targeted questioning from students and drive engagement, AMA is a specialized chatbot that answers only topic--specific questions in three areas--astronomy, sneakers and shoes, and dinosaurs. We tested AMA with 63 students in a K-8 public school in the Northeast USA. Students worked in small groups, interacted with our tool for three to ten minutes, and completed a post survey. We identified three key themes that emerged from student conversational interactions with AMA: expressing wonder, surprise, and curiosity; building trust and developing confidence; and building relationships and anthropomorphizing. Also, we observed a broad attitude of openness and comfort. Students trusted the chatbot responses in general, indicating a high level of trust in and reliance on AI as a source of information. They described AMA as "knowledgeable," "smart," and that the

For over 165 million years, dinosaurs reigned on this planet. Their entire existence saw variations in their body size and mass . Understanding the relationship between various attributes such as femur length, breadth; humerus length, breadth; tibia length, breadth and body mass of dinosaurs contributes to our understanding of the Jurassic era and further provides reasoning for bone and body size evolution of modern day descendants of those from the Dinosauria clade. The following work consists of statistical evidence derived from an encyclopedic data set consisting of a wide variety of measurements pertaining to discovered fossils of a particular taxa of dinosaur. Our study establishes linearly regressive correspondence between femur and humerus length and radii. Furthermore, there is also a comparison with terrestrial bird bone lengths, to verify the claim of birds being closest alive species to dinosaurs. An analysis into bone ratios of early humans shows that terrestrial birds are closer to humans than that of dinosaurs. Not only on one hand it challenges the closeness of birds with dinosaurs, but on the other hand it makes a case of convergent evolution between birds and human

Waveguide-coupled photonic crystal cavities with a triangular cross section fabricated by angled etching are suitable to interface embedded color centers with flying photonic qubits in quantum information applications. Moreover, their fabrication requires fewer processing steps compared to nanostructures produced by quasi-isotropic undercutting. As an alternative to established hole-based photonic crystal cavities, we introduce corrugated triangular 'Dinosaur' photonic crystal cavities, and develop a tapered, quasi loss-free cavity-waveguide interface to adiabatically interconvert Bloch and waveguide modes. We optimize the cavity-waveguide interface to minimize photon losses and demonstrate that its adjustment allows precise tuning of the light-matter interaction.

Spin-active colour centres in 4H silicon carbide are promising candidates as building blocks for quantum information applications. To increase the photon count rate of the emitters at low temperatures, the colour centres must be integrated into nanophotonic structures and characterised under cryogenic conditions. Here, we design and fabricate waveguide structures attached with an efficient Dinosaur photonic crystal reflector at one side. The devices show broadband reflection over a range of 60 THz with a peak reflectance above 80 %. Additionally, colour centres were integrated into these structures and characterised at cryogenic conditions. The emission was collected by a tapered-waveguide-tapered-fibre interface. Although the spectral stability of the emitters must be further improved for high excitation powers, the saturation intensity in standard PLE measurements is about 104 kcps. The count rate can be further improved to about 125 kcps with a charge-resonance check measurement scheme. To highlight the relevance of our devices, we theoretically show that these count rates enable optical single-shot readout with a fidelity exceeding 98 %.

Unsupervised multi-object discovery (MOD) aims to detect and localize distinct object instances in visual scenes without any form of human supervision. Recent approaches leverage object-centric learning (OCL) and motion cues from video to identify individual objects. However, these approaches use supervision to generate pseudo labels to train the OCL model. We address this limitation with MR-DINOSAUR -- Motion-Refined DINOSAUR -- a minimalistic unsupervised approach that extends the self-supervised pre-trained OCL model, DINOSAUR, to the task of unsupervised multi-object discovery. We generate high-quality unsupervised pseudo labels by retrieving video frames without camera motion for which we perform motion segmentation of unsupervised optical flow. We refine DINOSAUR's slot representations using these pseudo labels and train a slot deactivation module to assign slots to foreground and background. Despite its conceptual simplicity, MR-DINOSAUR achieves strong multi-object discovery results on the TRI-PD and KITTI datasets, outperforming the previous state of the art despite being fully unsupervised.

Lampreys are one of the oldest species in the world, living longer than dinosaurs, which is related to the ability to change the sex ratio during their lifespan. In this paper, to understand how sex ratio and food quantity affect the population growth rate of lampreys, the researchers draw inspiration from the logistics model and established a model called EcoSexChange(ESC), which results in a population initially increasing and then stabilizing, a reasonable outcome that may apply to other organisms with significant differences in consumption between sexes. Subsequently, this paper develops the Sex Ratio Adaptation Eco Impact (SRAEI) model based on the ESC model using the ABM algorithm to simulate how the population of lampreys, whose lives are divided into seven stages, grows and stabilizes. Then introduces a sudden disaster factor in the middle of the simulation, while also comparing lampreys that cannot adjust their sex ratio. The results of this paper are of great reference significance for people to analyze the population changes of lampreys in different living environments, and they are also easy to apply to other species with large differences between males and females.