Detection of unwanted (`foreign') objects within products is a common procedure in many branches of industry for maintaining production quality. X-ray imaging is a fast, non-invasive and widely applicable method for foreign object detection. Deep learning has recently emerged as a powerful approach for recognizing patterns in radiographs (i.e., X-ray images), enabling automated X-ray based foreign object detection. However, these methods require a large number of training examples and manual annotation of these examples is a subjective and laborious task. In this work, we propose a Computed Tomography (CT) based method for producing training data for supervised learning of foreign object detection, with minimal labour requirements. In our approach, a few representative objects are CT scanned and reconstructed in 3D. The radiographs that have been acquired as part of the CT-scan data serve as input for the machine learning method. High-quality ground truth locations of the foreign objects are obtained through accurate 3D reconstructions and segmentations. Using these segmented volumes, corresponding 2D segmentations are obtained by creating virtual projections. We outline the benefits of objectively and reproducibly generating training data in this way compared to conventional radiograph annotation. In addition, we show how the accuracy depends on the number of objects used for the CT reconstructions. The results show that in this workflow generally only a relatively small number of representative objects (i.e., fewer than 10) are needed to achieve adequate detection performance in an industrial setting. Moreover, for real experimental data we show that the workflow leads to higher foreign object detection accuracies than with standard radiograph annotation.

You can check the schedule in your city, here. Description This session aims to reveal the infrastructure and energy resources that are needed to support digitization, massive data processing and most Artificial Intelligence developments, emphasizing the ecological footprint of its processes. Every week you will receive the links for the sessions, book it on your calendar!

Many astrophysical analyses depend on estimates of redshifts (a proxy for distance) determined from photometric (i.e., imaging) data alone. Inaccurate estimates of photometric redshift uncertainties can result in large systematic errors. However, probability distribution outputs from many photometric redshift methods do not follow the frequentist definition of a Probability Density Function (PDF) for redshift -- i.e., the fraction of times the true redshift falls between two limits $z_{1}$ and $z_{2}$ should be equal to the integral of the PDF between these limits. Previous works have used the global distribution of Probability Integral Transform (PIT) values to re-calibrate PDFs, but offsetting inaccuracies in different regions of feature space can conspire to limit the efficacy of the method. We leverage a recently developed regression technique that characterizes the local PIT distribution at any location in feature space to perform a local re-calibration of photometric redshift PDFs. Though we focus on an example from astrophysics, our method can produce PDFs which are calibrated at all locations in feature space for any use case.

We benchmark the current existing methods to jointly learn non-Cartesian k-space trajectory and reconstruction: PILOT, BJORK, and compare them with those obtained from the recently developed generalized hybrid learning (HybLearn) framework. We present the advantages of using projected gradient descent to enforce MR scanner hardware constraints as compared to using added penalties in the cost function. Further, we use the novel HybLearn scheme to jointly learn and compare our results through a retrospective study on fastMRI validation dataset.

Recent approaches of computer vision utilize deep learning methods as they perform quite well if training and testing domains follow the same underlying data distribution. However, it has been shown that minor variations in the images that occur when using these methods in the real world can lead to unpredictable errors. Transfer learning is the area of machine learning that tries to prevent these errors. Especially, approaches that augment image data using auxiliary knowledge encoded in language embeddings or knowledge graphs (KGs) have achieved promising results in recent years. This survey focuses on visual transfer learning approaches using KGs. KGs can represent auxiliary knowledge either in an underlying graph-structured schema or in a vector-based knowledge graph embedding. Intending to enable the reader to solve visual transfer learning problems with the help of specific KG-DL configurations we start with a description of relevant modeling structures of a KG of various expressions, such as directed labeled graphs, hypergraphs, and hyper-relational graphs. We explain the notion of feature extractor, while specifically referring to visual and semantic features. We provide a broad overview of knowledge graph embedding methods and describe several joint training objectives suitable to combine them with high dimensional visual embeddings. The main section introduces four different categories on how a KG can be combined with a DL pipeline: 1) Knowledge Graph as a Reviewer; 2) Knowledge Graph as a Trainee; 3) Knowledge Graph as a Trainer; and 4) Knowledge Graph as a Peer. To help researchers find evaluation benchmarks, we provide an overview of generic KGs and a set of image processing datasets and benchmarks including various types of auxiliary knowledge. Last, we summarize related surveys and give an outlook about challenges and open issues for future research.

Compositional generalization is a troubling blind spot for neural language models. Recent efforts have presented techniques for improving a model's ability to encode novel combinations of known inputs, but less work has focused on generating novel combinations of known outputs. Here we focus on this latter "decode-side" form of generalization in the context of gSCAN, a synthetic benchmark for compositional generalization in grounded language understanding. We present Recursive Decoding (RD), a novel procedure for training and using seq2seq models, targeted towards decode-side generalization. Rather than generating an entire output sequence in one pass, models are trained to predict one token at a time. Inputs (i.e., the external gSCAN environment) are then incrementally updated based on predicted tokens, and re-encoded for the next decoder time step. RD thus decomposes a complex, out-of-distribution sequence generation task into a series of incremental predictions that each resemble what the model has already seen during training. RD yields dramatic improvement on two previously neglected generalization tasks in gSCAN. We provide analyses to elucidate these gains over failure of a baseline, and then discuss implications for generalization in naturalistic grounded language understanding, and seq2seq more generally.

Ensuring public safety in a Smart City (SC) environment is a critical and increasingly complicated task due to the involvement of multiple agencies and the city's expansion across cyber and social layers. In this paper, we propose an extensive form perfect information game to model interactions and optimal city resource allocations when a Terrorist Organization (TO) performs attacks on multiple targets across two conceptual SC levels, a physical, and a cyber-social. The Smart City Defense Game (SCDG) considers three players that initially are entitled to a specific finite budget. Two SC agencies that have to defend their physical or social territories respectively, fight against a common enemy, the TO. Each layer consists of multiple targets and the attack outcome depends on whether the resources allocated there by the associated agency, exceed or not the TO's. Each player's utility is equal to the number of successfully defended targets. The two agencies are allowed to make budget transfers provided that it is beneficial for both. We completely characterize the Sub-game Perfect Nash Equilibrium (SPNE) of the SCDG that consists of strategies for optimal resource exchanges between SC agencies and accounts for the TO's budget allocation across the physical and social targets. Also, we present numerical and comparative results demonstrating that when the SC players act according to the SPNE, they maximize the number of successfully defended targets. The SCDG is shown to be a promising solution for modeling critical resource allocations between SC parties in the face of multi-layer simultaneous terrorist attacks.

One of the most important and challenging problems in football is predicting future player performance when transferred to another club within and between different leagues. In addition to being the most valuable prediction a team makes, it is also the most complex analytics task to perform as it needs to take into consideration: a) differences in playing style between the player's current team and target team, b) differences in style and ability of other players on each team, c) differences in league quality and style, and d) the role the player is desired to play. In this paper, we present a method which addresses these issues and enables us to make accurate predictions of future performance. Our Transfer Portal model utilizes a personalized neural network accounting for both stylistic and ability level input representations for players, teams, and leagues to simulate future player performance at any chosen club. Furthermore, we use a Bayesian updating framework to dynamically modify player and team representations over time which enables us to generate predictions for rising stars with small amounts of data.

Around 2015 or 2016, Google began a strategic move, possibly one of the most important it has undertaken in its more than 20 years of existence, to incorporate machine learning and apply it to all its products. Its CEO, Sundar Pichai, saw machine learning as the ultimate disruptive technology, in the same way fire or electricity once were, and in a strong display of leadership, embarked on a mission to introduce the technology throughout the company, training all its employees in it. A few years on, what can we observe? I think email, a tool we all use quite regularly, is an interesting example. What changes have we noticed over the last few years with Gmail, for example?

Microsoft continues to make significant investments in deep learning, computer vision, and AI. The Microsoft Maps Team has been leveraging that investment to identify map features at scale and produce high-quality building footprint data sets with the overall goal to add to the OpenStreetMap and MissingMaps humanitarian efforts. As of this post, the following locations are available and Microsoft offers access to this data under the Open Data Commons Open Database License (ODbL). Country/Region Million buildings United States of America 129.6 Nigeria and Kenya 50.5 South America 44.5 Uganda and Tanzania 17.9 Canada 11.8 Australia 11.3 As you might expect, the vintage of the footprints depends on the collection date of the underlying imagery. Bing Maps Imagery is a composite of multiple sources with different capture dates (ranging 2012 to 2021).