Ensuring reliability is paramount in deep learning, particularly within the domain of medical imaging, where diagnostic decisions often hinge on model outputs. The capacity to separate out-of-distribution (OOD) samples has proven to be a valuable indicator of a model's reliability in research. In medical imaging, this is especially critical, as identifying OOD inputs can help flag potential anomalies that might otherwise go undetected. While many OOD detection methods rely on feature or logit space representations, recent works suggest these approaches may not fully capture OOD diversity. To address this, we propose a novel OOD scoring mechanism, called NERO, that leverages neuron-level relevance at the feature layer. Specifically, we cluster neuron-level relevance for each in-distribution (ID) class to form representative centroids and introduce a relevance distance metric to quantify a new sample's deviation from these centroids, enhancing OOD separability. Additionally, we refine performance by incorporating scaled relevance in the bias term and combining feature norms. Our framework also enables explainable OOD detection.

Concept learning deals with learning description logic concepts from a background knowledge and input examples. The goal is to learn a concept that covers all positive examples, while not covering any negative examples. This non-trivial task is often formulated as a search problem within an infinite quasi-ordered concept space. Although state-of-the-art models have been successfully applied to tackle this problem, their large-scale applications have been severely hindered due to their excessive exploration incurring impractical runtimes. Here, we propose a remedy for this limitation. We reformulate the learning problem as a multi-label classification problem and propose a neural embedding model (NERO) that learns permutation-invariant embeddings for sets of examples tailored towards predicting $F_1$ scores of pre-selected description logic concepts. By ranking such concepts in descending order of predicted scores, a possible goal concept can be detected within few retrieval operations, i.e., no excessive exploration. Importantly, top-ranked concepts can be used to start the search procedure of state-of-the-art symbolic models in multiple advantageous regions of a concept space, rather than starting it in the most general concept $\top$. Our experiments on 5 benchmark datasets with 770 learning problems firmly suggest that NERO significantly (p-value <1%) outperforms the state-of-the-art models in terms of $F_1$ score, the number of explored concepts, and the total runtime. We provide an open-source implementation of our approach.

The days of having a low-quality image you found on the internet are over. Thanks to the magic of an image upscaler, you never have to settle for low-res again. If you're familiar with AI upscale, then you may have heard of Topaz Photo AI. The cost of Topaz Photo AI, unfortunately, leaves you having to shell out a lot of cash. Luckily, there are a lot of other excellent free alternatives!

Explaining the predictions of AI models is paramount in safety-critical applications, such as in legal or medical domains. One form of explanation for a prediction is an extractive rationale, i.e., a subset of features of an instance that lead the model to give its prediction on the instance. Previous works on generating extractive rationales usually employ a two-phase model: a selector that selects the most important features (i.e., the rationale) followed by a predictor that makes the prediction based exclusively on the selected features. One disadvantage of these works is that the main signal for learning to select features comes from the comparison of the answers given by the predictor and the ground-truth answers. In this work, we propose to squeeze more information from the predictor via an information calibration method. More precisely, we train two models jointly: one is a typical neural model that solves the task at hand in an accurate but black-box manner, and the other is a selector-predictor model that additionally produces a rationale for its prediction. The first model is used as a guide to the second model. We use an adversarial-based technique to calibrate the information extracted by the two models such that the difference between them is an indicator of the missed or over-selected features. In addition, for natural language tasks, we propose to use a language-model-based regularizer to encourage the extraction of fluent rationales. Experimental results on a sentiment analysis task as well as on three tasks from the legal domain show the effectiveness of our approach to rationale extraction.

The problem is when Devil May Cry is at its best, it defies explanation. Sure you can try, and I'm going to for professional reasons. But half the game is "You have to see it to believe it" and the other half is "You have to play it to understand" and the two meet at the middle (albeit joined by some clunky loading screens) to create a relentlessly entertaining video game--one that's self-confident, satisfying, and stylish as hell. Imagine my surprise, that in the span of a year Capcom could make me a fan of first Monster Hunter, then Resident Evil, and now Devil May Cry. It's an incredible run, by a company that a few years ago I would've said seemed listless. In any case, this is the first Devil May Cry in over a decade--for fans, that is.

Personalization has become integral to the customer journey and is now a key driver of brand loyalty across all channels. Consumers are much more likely to buy from brands – both in-store and online – when offers are personalized. And it's not just your brand communications that need to be more relevant: consumers are also interested in purchasing more personalized products and services, and are willing to wait longer to get them. You know more about your customers than ever before. But isn't one of your biggest challenges how to make sense of all that customer data so your marketing messages can be more targeted and relevant?

Ten years after the debut of Devil May Cry 4, Nero is back in the driver's seat and he's never looked better. It's not just the haircut, either -- Devil May Cry 5 runs on the RE Engine built for Resident Evil 7: Biohazard, and Capcom's goal is to build a photorealistic game filled with gooey demons, witty one-liners and flashy combos. The title's first hands-on demo at Gamescom 2018 highlights these exact elements and wraps all of it up in an ichor-crusted, gorgeous package. Devil May Cry 4 was the first game in the franchise to star Nero, a reluctant ally to series protagonist Dante. Nero is a human with a smattering of supernatural abilities, including a demon-powered arm named Devil Bringer.

Devil May Cry 5 is due to land on Xbox One, PlayStation 4 and PC on March 8th, 2019, Capcom announced today during the special Gamescom edition of the Inside Xbox livestream. Devil May Cry 5 is the sequel to Devil May Cry 4, which came out in 2008, and it marks the return of series shepherd Hideaki Itsuno as director. Ninja Theory briefly took over the franchise with DmC: Devil May Cry in 2013, and while Devil May Cry 5 takes some styling tips from that game, it doesn't follow its storyline directly. Devil May Cry 5 picks up where Devil May Cry 4 left off, and it stars three playable characters: Dante, Nero and an unknown protagonist in a long black trench vest. Nero historically has a demon-powered arm called Devil Bringer, and it's ripped from his body at the beginning of Devil May Cry 5. It's replaced by a series of robotic arms (called Devil Breakers) created by Nero's partner in demon-fighting crime, Nico.