Search

Unsupervised learning of keypoints and landmarks has seen significant progress with the help of modern neural network architectures, but performance is yet to match the supervised counterpart, making their practicability questionable. We leverage the emergent knowledge within text-to-image diffusion models, towards more robust unsupervised keypoints. Our core idea is to find text embeddings that would cause the generative model to consistently attend to compact regions in images (i.e. keypoints). To do so, we simply optimize the text embedding such that the cross-attention maps within the denoising network are localized as Gaussians with small standard deviations. We validate our performance on multiple datasets: the CelebA, CUB-200-2011, Tai-Chi-HD, DeepFashion, and Human3.6m datasets. We achieve significantly improved accuracy, sometimes even outperforming supervised ones, particularly for data that is non-aligned and less curated. Our code is publicly available and can be found through our project page: https://ubc-vision.github.io/StableKeypoints/

Most virtual try-on research is motivated to serve the fashion business by generating images to demonstrate garments on studio models at a lower cost. However, virtual try-on should be a broader application that also allows customers to visualize garments on themselves using their own casual photos, known as in-the-wild try-on. Unfortunately, the existing methods, which achieve plausible results for studio try-on settings, perform poorly in the in-the-wild context. This is because these methods often require paired images (garment images paired with images of people wearing the same garment) for training. While such paired data is easy to collect from shopping websites for studio settings, it is difficult to obtain for in-the-wild scenes. In this work, we fill the gap by (1) introducing a StreetTryOn benchmark to support in-the-wild virtual try-on applications and (2) proposing a novel method to learn virtual try-on from a set of in-the-wild person images directly without requiring paired data. We tackle the unique challenges, including warping garments to more diverse human poses and rendering more complex backgrounds faithfully, by a novel DensePose warping correction method combined with diffusion-based conditional inpainting. Our experiments show competitive performance for standard studio try-on tasks and SOTA performance for street try-on and cross-domain try-on tasks., Comment: The abstract and intro are updated. Some typos and some pdf rendering errors have been fixed in the version

With the increasingly widespread adoption of AI in healthcare, maintaining the accuracy and reliability of AI models in clinical practice has become crucial. In this context, we introduce novel methods for monitoring the performance of radiology AI classification models in practice, addressing the challenges of obtaining real-time ground truth for performance monitoring. We propose two metrics - predictive divergence and temporal stability - to be used for preemptive alerts of AI performance changes. Predictive divergence, measured using Kullback-Leibler and Jensen-Shannon divergences, evaluates model accuracy by comparing predictions with those of two supplementary models. Temporal stability is assessed through a comparison of current predictions against historical moving averages, identifying potential model decay or data drift. This approach was retrospectively validated using chest X-ray data from a single-center imaging clinic, demonstrating its effectiveness in maintaining AI model reliability. By providing continuous, real-time insights into model performance, our system ensures the safe and effective use of AI in clinical decision-making, paving the way for more robust AI integration in healthcare, Comment: 10 pages, 4 figures, 2 tables

We study the photo-physical stability of ensemble near-surface nitrogen vacancy (NV) centers in diamond under vacuum and air. The optically detected magnetic resonance contrast of the NV centers was measured following exposure to laser illumination, showing opposing trends in air compared to vacuum (increasing by up to 9% and dropping by up to 25%, respectively). Characterization using Raman and X-ray photoelectron spectroscopies suggests a surface reconstruction: In air, atmospheric oxygen adsorption on surface leads to an increase in NV- fraction, whereas in vacuum, net oxygen desorption increases the NV0 fraction. NV charge state switching is confirmed by photoluminescence spectroscopy. Deposition of ~ 2 nm alumina (Al2O3) over the diamond surface was shown to stabilize the NV charge state under illumination in either environment, attributed to a more stable surface electronegativity. The use of alumina coating on diamond is therefore a promising approach to improve the resilience of NV sensors.

The observed scaling relations between supermassive black hole masses and their host galaxy properties indicate that supermassive black holes influence the evolution of galaxies. However, the scaling relations may be affected by selection biases. We propose to measure black hole masses in a mass-limited galaxy sample including all non-detections to inprove constraints on galaxy mass - black hole mass scaling relations and test for selection bias. We use high spatial resolution spectroscopy from the Keck and Gemini telescopes, and the Jeans Anisotropic Modelling method to measure black hole masses in early type galaxies from the Virgo Cluster. We present four new black hole masses and one upper limit in our mass-selected sample of galaxies of galaxy mass (1.0-3.2) x $10^{10} M_\odot$. This brings the total measured to 11 galaxies out of a full sample of 18 galaxies, allowing us to constrain scaling relations. We calculate a lower limit for the average black hole mass in our sample of 3.7 x $10^{7} M_\odot$. This is at an average galaxy stellar mass of (1.81 +/- 0.14) x $10^{10} M_\odot $ and an average bulge mass of (1.31 +/- 0.15) x $10^{10} M_\odot$. This lower limit shows that black hole masses in early type galaxies are not strongly affected by selection biases.

Orca 1 learns from rich signals, such as explanation traces, allowing it to outperform conventional instruction-tuned models on benchmarks like BigBench Hard and AGIEval. In Orca 2, we continue exploring how improved training signals can enhance smaller LMs' reasoning abilities. Research on training small LMs has often relied on imitation learning to replicate the output of more capable models. We contend that excessive emphasis on imitation may restrict the potential of smaller models. We seek to teach small LMs to employ different solution strategies for different tasks, potentially different from the one used by the larger model. For example, while larger models might provide a direct answer to a complex task, smaller models may not have the same capacity. In Orca 2, we teach the model various reasoning techniques (step-by-step, recall then generate, recall-reason-generate, direct answer, etc.). More crucially, we aim to help the model learn to determine the most effective solution strategy for each task. We evaluate Orca 2 using a comprehensive set of 15 diverse benchmarks (corresponding to approximately 100 tasks and over 36,000 unique prompts). Orca 2 significantly surpasses models of similar size and attains performance levels similar or better to those of models 5-10x larger, as assessed on complex tasks that test advanced reasoning abilities in zero-shot settings. make Orca 2 weights publicly available at aka.ms/orca-lm to support research on the development, evaluation, and alignment of smaller LMs, Comment: Added url to model weights fixed typo in Author name

We introduce Style Tailoring, a recipe to finetune Latent Diffusion Models (LDMs) in a distinct domain with high visual quality, prompt alignment and scene diversity. We choose sticker image generation as the target domain, as the images significantly differ from photorealistic samples typically generated by large-scale LDMs. We start with a competent text-to-image model, like Emu, and show that relying on prompt engineering with a photorealistic model to generate stickers leads to poor prompt alignment and scene diversity. To overcome these drawbacks, we first finetune Emu on millions of sticker-like images collected using weak supervision to elicit diversity. Next, we curate human-in-the-loop (HITL) Alignment and Style datasets from model generations, and finetune to improve prompt alignment and style alignment respectively. Sequential finetuning on these datasets poses a tradeoff between better style alignment and prompt alignment gains. To address this tradeoff, we propose a novel fine-tuning method called Style Tailoring, which jointly fits the content and style distribution and achieves best tradeoff. Evaluation results show our method improves visual quality by 14%, prompt alignment by 16.2% and scene diversity by 15.3%, compared to prompt engineering the base Emu model for stickers generation., Comment: 10 pages, 5 figures

From the humble beginnings of particle physics to groundbreaking discoveries, the journey of particle physics has revolutionized our understanding of the universe. The progression of particle physics, begin from the discovery of new particles, to gauge symmetries and many successful theories in its journey which help in building up the model that explain the underlying makeup of matter, regulated by fundamental forces led to establishment of Standard Model(SM). The discovery of neutrino oscillations compelled scientists to explore theories that extend beyond the Standard Model, which opened up new frontiers in the field of particle physics. The theories have shifted its paradigm to a single comprehensive theory at the scale of 10^14 GeV. This article explains the journey of particle physics from the basic constituent to the theory of everything., Comment: 35 pages,200 refrences

We present a comparison of the presence and properties of dust in two distinct phases of the Milky Way's interstellar medium: the warm neutral medium (WNM) and the warm ionized medium (WIM). Using distant pulsars at high Galactic latitudes and vertical distance ($|b| > 40\deg$, $D \sin|b| > 2 \mathrm{\,\, kpc}$) as probes, we measure their dispersion measures and the neutral hydrogen component of the warm neutral medium ($\text{WNM}_\text{HI}$) using HI column density. Together with dust intensity along these same sightlines, we separate the respective dust contributions of each ISM phase in order to determine whether the ionized component contributes to the dust signal. We measure the temperature ($T$), spectral index ($\beta$), and dust opacity ($\tau/N_{H}$) in both phases. We find $T~{\text{(WNM}_\text{HI})}=20^{+3}_{-2}$~K, $\beta~{\text{(WNM}_\text{HI})} = 1.5\pm{0.4}$, and $\tau_{\text{353}}/N_{H}~{\text{(WNM}_\text{HI})}=(1.0\pm0.1)\times 10^{-26}$~cm$^2$. Assuming that the temperature and spectral index are the same in both the WNM$_\text{HI}$ and WIM, and given our simple model that widely separated lines-of-sight can be fit together, we find evidence that there is a dust signal associated with the ionized gas and $\tau_{\text{353}}/N_{H}~\text{(WIM)}=(0.3\pm0.3)\times 10^{-26}$, which is about three times smaller than $\tau_{\text{353}}/N_{H}~{\text{(WNM}_\text{HI})}$. We are 80% confident that $\tau_{\text{353}}/N_{H}~\text{(WIM)}$ is at least two times smaller than $\tau_{\text{353}}/N_{H}~{\text{(WNM}_\text{HI})}$., Comment: 11 pages, 4 tables, 4 figures, accepted to the Astrophysical Journal

We introduce and study the problem of adversarial arithmetic, which provides a simple yet challenging testbed for language model alignment. This problem is comprised of arithmetic questions posed in natural language, with an arbitrary adversarial string inserted before the question is complete. Even in the simple setting of 1-digit addition problems, it is easy to find adversarial prompts that make all tested models (including PaLM2, GPT4, Claude2) misbehave, and even to steer models to a particular wrong answer. We additionally provide a simple algorithm for finding successful attacks by querying those same models, which we name "prompt inversion rejection sampling" (PIRS). We finally show that models can be partially hardened against these attacks via reinforcement learning and via agentic constitutional loops. However, we were not able to make a language model fully robust against adversarial arithmetic attacks.

We compute instanton corrections to the partition function of sine-Liouville (SL) theory, which provides a worldsheet description of two-dimensional string theory in a non-trivial tachyon background. We derive these corrections using a matrix model formulation based on a chiral representation of matrix quantum mechanics and using string theory methods. In both cases we restrict to the leading and subleading orders in the string coupling expansion. Then the CFT technique is used to compute two orders of the expansion in the SL perturbation parameter $\lambda$, while the matrix model gives results which are non-perturbative in $\lambda$. The matrix model results perfectly match those of string theory in the small $\lambda$ expansion. We also generalize our findings to the case of perturbation by several tachyon vertex operators carrying different momenta, and obtain interesting analytic predictions for the disk two-point and annulus one-point functions with ZZ boundary condition., Comment: 47+30 pages

In this work, we study a new hollow-core (air-filled) double-clad anti-resonant fiber (DC-ARF) as a potent candidate for multiphoton micro-endoscopy. We compare the fiber characteristics with a single-clad anti-resonant fiber (SC-ARF) and a solid core fiber (SCF). While the DC-ARF and the SC-ARF enable low-loss (<0.2 dBm-1), close to dispersion-free excitation pulse delivery (<10% pulse width increase at 900 nm per 1 m fiber) without any induced non-linearities, the SCF resulted in spectral broadening and pulse-stretching (> 2000% of pulse width increase at 900 nm per 1 m fiber). An ideal optical fiber endoscope needs to be several meters long and should enable both excitation and collection through the fiber. Therefore, we performed multiphoton imaging on endoscopy-compatible 1 m and 3 m lengths of fiber in the back-scattered geometry, wherein the signals were collected either directly (non-descanned detection) or through the fiber (descanned detection). Second harmonic images were collected from barium titanate crystals as well as from biological samples (rat tail tendon). In non-descanned detection conditions, the ARFs outperformed the SCF by up to 10 times in terms of signal-to-noise ratio of images. Significantly, only the DC-ARF, due to its high numerical aperture (0.45) and wide-collection bandwidth (>1 um), could provide images in the de-scanned detection configuration desirable for endoscopy. Thus, our systematic characterization and comparison of different optical fibres under different image collection configurations, confirms and establishes the utility of DC-ARFs for high-performing label-free multiphoton imaging based micro-endoscopy., Comment: 29 pages, 13 figures

A fundamental dynamical constraint -- that fluctuation induced charge-weighted particle flux must vanish -- can prevent instabilities from accessing the free energy in the strong gradients characteristic of Transport Barriers (TBs). Density gradients, when larger than a certain threshold, lead to a violation of the constraint and emerge as a stabilizing force. This mechanism, then, broadens the class of configurations (in magnetized plasmas) where these high confinement states can be formed and sustained. The need for velocity shear, the conventional agent for TB formation, is obviated. The most important ramifications of the constraint is to permit a charting out of the domains conducive to TB formation and hence to optimally confined fusion worthy states; the detailed investigation is conducted through new analytic methods and extensive gyrokinetic simulations.

In this paper, we present a model of a game among teams. Each team consists of a homogeneous population of agents. Agents within a team are cooperative while the teams compete with other teams. The dynamics and the costs are coupled through the empirical distribution (or the mean field) of the state of agents in each team. This mean-field is assumed to be observed by all agents. Agents have asymmetric information (also called a non-classical information structure). We propose a mean-field based refinement of the Team-Nash equilibrium of the game, which we call mean-field Markov perfect equilibrium (MF-MPE). We identify a dynamic programming decomposition to characterize MF-MPE. We then consider the case where each team has a large number of players and present a mean-field approximation which approximates the game among large-population teams as a game among infinite-population teams. We show that MF-MPE of the game among teams of infinite population is easier to compute and is an $\varepsilon$-approximate MF-MPE of the game among teams of finite population., Comment: 20 pages

The free energy balance equation for gyrokinetic fluctuations is derived and applied to instabilities. An additional term due to electromagnetic sources is included. This can provide a simpler way to compute the free energy balance in practical applications, and is also conceptually clarifying. The free energy balance, by itself, is not sufficient to determine an eigenfrequency. The preceding results are derived in general geometry. The charge flux constraint in gyrokinetics can provide a necessary additional relation, and the combination of these two can be equivalent to a dispersion relation. The charge flux constraint can prevent the appearance of an unstable eigenmode even though the free energy balance would allow strongly growing fluctuations. The application of these concepts to simplified kinetic models in simplified geometry is also indicated., Comment: 5 pages

In this paper, the numerical approximation of the generalized Burgers'-Huxley equation (GBHE) with weakly singular kernels using non-conforming methods will be presented. Specifically, we discuss two new formulations. The first formulation is based on the non-conforming finite element method (NCFEM). The other formulation is based on discontinuous Galerkin finite element methods (DGFEM). The wellposedness results for both formulations are proved. Then, a priori error estimates for both the semi-discrete and fully-discrete schemes are derived. Specific numerical examples, including some applications for the GBHE with weakly singular model, are discussed to validate the theoretical results.

Based on the recently demonstrated resonant wave-wave process, it is shown that electrons can be accelerated to ultra-relativistic energies in the magnetospheres of radio pulsars. The energization occurs via the resonant interaction of the electron wave (described by a Klein-Gordon (KG) equation) moving in unison with an intense electromagnetic (EM) wave; the KG wave/particle continuously draws energy from EM. In a brief recapitulation of the general theory, the high energy (resonantly enhanced) electron states are investigated by solving the KG equation, minimally coupled to the EM field. The restricted class of solutions, that propagate in phase with EM radiation (functions only of $\zeta=\omega t-kz$), are explored to serve as a possible basis for the proposed electron energization in the radio pulsars. We show that the wave-wave resonant energization mechanism could be operative in a broad class of radio pulsars with periods ranging from milliseconds to the normal values ($\sim 1$ sec); it could drive the magnetospheric electrons to acquire energies from $100$s of TeVs (millisecond pulsars) to $10$ ZeVs (normal pulsars)., Comment: 5 pages, 2 figures