While Fourier-based neural operators are best suited to learning mappings between functions on periodic domains, several works have introduced techniques for incorporating non trivial boundary conditions. However, all previously introduced methods have restrictions that limit their applicability. In this work, we introduce an alternative approach to imposing boundary conditions inspired by volume penalization from numerical methods and Mixture of Experts (MoE) from machine learning. By introducing competing experts, the approach additionally allows for model selection. To demonstrate the method, we combine a spatially conditioned MoE with the Fourier based, Modal Operator Regression for Physics (MOR-Physics) neural operator and recover a nonlinear operator on a disk and quarter disk. Next, we extract a large eddy simulation (LES) model from direct numerical simulation of channel flow and show the domain decomposition provided by our approach. Finally, we train our LES model with Bayesian variational inference and obtain posterior predictive samples of flow far past the DNS simulation time horizon.

Cryptocurrencies are highly volatile financial instruments with more and more new retail investors joining the scene with each passing day. Bitcoin has always proved to determine in which way the rest of the cryptocurrency market is headed towards. As of today Bitcoin has a market dominance of close to 50 percent. Bull and bear phases in cryptocurrencies are determined based on the performance of Bitcoin over the 50 Day and 200 Day Moving Averages. The aim of this paper is to foretell the performance of bitcoin in the near future by employing predictive algorithms. This predicted data will then be used to calculate the 50 Day and 200 Day Moving Averages and subsequently plotted to establish the potential bull and bear phases.

We present SSL-HV: Self-Supervised Learning approaches applied to the task of Handwriting Verification. This task involves determining whether a given pair of handwritten images originate from the same or different writer distribution. We have compared the performance of multiple generative, contrastive SSL approaches against handcrafted feature extractors and supervised learning on CEDAR AND dataset. We show that ResNet based Variational Auto-Encoder (VAE) outperforms other generative approaches achieving 76.3% accuracy, while ResNet-18 fine-tuned using Variance-Invariance-Covariance Regularization (VICReg) outperforms other contrastive approaches achieving 78% accuracy. Using a pre-trained VAE and VICReg for the downstream task of writer verification we observed a relative improvement in accuracy of 6.7% and 9% over ResNet-18 supervised baseline with 10% writer labels.

Apple, in its latest update to its App Store developer guidelines for iPhones and iPads, flagged by 9to5Mac, says it will allow game console emulators – and even downloadable games. Apple warns developers, however, they "are responsible for all such software offered in [their] app, including ensuring that such software complies with these Guidelines and all applicable laws." So don't expect to play Super Mario, Spyro, or a third game series that starts with an'S'. Meanwhile, we have a guide to watching (and recording) the total eclipse in North America later today. The best chance of good viewing along the path of eclipse totality is still in northeastern parts of the US (Buffalo, NY, Burlington, VT) and southeast Canada (Niagara Falls and Montreal).

A group of Tesla workers in New York has sent company chief Elon Musk a letter stating their intention to unionize, according to Bloomberg. It could end up being the first Tesla union if successful, seeing as previous attempts fizzled out before organizers could petition for a vote. The employees involved in the campaign are in charge of labeling data for Tesla's Autopilot technology at the company's Buffalo, New York facility. Bloomberg says the group is asking for better pay, job security and a better work environment that eases the production pressures placed on them. Workers told the news organization that they've been skipping bathroom breaks, since Tesla keeps a close eye on their every move.

Maura R. Grossman, JD, Ph.D., is a Research Professor in the Cheriton School of Computer Science, an Adjunct Professor at Osgoode Hall Law School, and an affiliate faculty member of the Vector Institute for Artificial Intelligence. She is also Principal at Maura Grossman Law, an eDiscovery law and consulting firm in Buffalo, New York. Maura is best known for her work on technology-assisted review, a supervised machine learning approach that she and her colleague, Computer Science Professor Gordon V. Cormack, developed to expedite review of documents in high-stakes litigation. She teaches Artificial Intelligence: Law, Ethics, and Policy, a course for graduate computer science students at Waterloo and upper-class law students at Osgoode, as well as the ethics workshop required of all students in the master's programs in artificial intelligence and data science at Waterloo. Artificial intelligence is an umbrella term first used at a conference in Dartmouth in 1956.

Pre-training visual and textual representations from large-scale image-text pairs is becoming a standard approach for many downstream vision-language tasks. The transformer-based models learn inter and intra-modal attention through a list of self-supervised learning tasks. This paper proposes LAViTeR, a novel architecture for visual and textual representation learning. The main module, Visual Textual Alignment (VTA) will be assisted by two auxiliary tasks, GAN-based image synthesis and Image Captioning. We also propose a new evaluation metric measuring the similarity between the learnt visual and textual embedding. The experimental results on two public datasets, CUB and MS-COCO, demonstrate superior visual and textual representation alignment in the joint feature embedding space

Self-supervised learning provides an opportunity to explore unlabeled chest X-rays and their associated free-text reports accumulated in clinical routine without manual supervision. This paper proposes a Joint Image Text Representation Learning Network (JoImTeRNet) for pre-training on chest X-ray images and their radiology reports. The model was pre-trained on both the global image-sentence level and the local image region-word level for visual-textual matching. Both are bidirectionally constrained on Cross-Entropy based and ranking-based Triplet Matching Losses. The region-word matching is calculated using the attention mechanism without direct supervision about their mapping. The pre-trained multi-modal representation learning paves the way for downstream tasks concerning image and/or text encoding. We demonstrate the representation learning quality by cross-modality retrievals and multilabel classifications on two datasets: OpenI-IU and MIMIC-CXR.

In forming learning objectives, one oftentimes needs to aggregate a set of individual values to a single output. Such cases occur in the aggregate loss, which combines individual losses of a learning model over each training sample, and in the individual loss for multi-label learning, which combines prediction scores over all class labels. In this work, we introduce the sum of ranked range (SoRR) as a general approach to form learning objectives. A ranked range is a consecutive sequence of sorted values of a set of real numbers. The minimization of SoRR is solved with the difference of convex algorithm (DCA). We explore two applications in machine learning of the minimization of the SoRR framework, namely the AoRR aggregate loss for binary/multi-class classification at the sample level and the TKML individual loss for multi-label/multi-class classification at the label level. A combination loss of AoRR and TKML is proposed as a new learning objective for improving the robustness of multi-label learning in the face of outliers in sample and labels alike. Our empirical results highlight the effectiveness of the proposed optimization frameworks and demonstrate the applicability of proposed losses using synthetic and real data sets.

Moves have been made to restrict the use of facial recognition across the globe. In part one of this series on Face ID, Jennifer Strong and the team at MIT Technology Review explore the unexpected ways the technology is being used, including how technology is being turned on police. This episode was reported and produced by Jennifer Strong, Tate Ryan-Mosley and Emma Cillekens, and Karen Hao. Strong: A few things have happened since we last spoke about facial recognition. We've seen more places move to restrict its use while at the same time, schools and other public buildings have started using face I-D as part of their covid-prevention plans. We're even using it on animals and not just on faces with similarities to our own, like chimps and gorillas, Chinese tech firms use it on pigs, and Canadian scientists are working to identify whales, even grizzly bears.