Failure detection in automated image classification is a critical safeguard for clinical deployment. Detected failure cases can be referred to human assessment, ensuring patient safety in computer-aided clinical decision making. Despite its paramount importance, there is insufficient evidence about the ability of state-of-the-art confidence scoring methods to detect test-time failures of classification models in the context of medical imaging. This paper provides a reality check, establishing the performance of in-domain misclassification detection methods, benchmarking 9 widely used confidence scores on 6 medical imaging datasets with different imaging modalities, in multiclass and binary classification settings. Our experiments show that the problem of failure detection is far from being solved. We found that none of the benchmarked advanced methods proposed in the computer vision and machine learning literature can consistently outperform a simple softmax baseline, demonstrating that improved out-of-distribution detection or model calibration do not necessarily translate to improved in-domain misclassification detection.

Recognizing an image and segmenting it into coherent regions are often treated as separate tasks. Human vision, however, has a general sense of segmentation hierarchy before recognition occurs. We are thus inspired to learn image recognition with hierarchical image segmentation based entirely on unlabeled images. Our insight is to learn fine-to-coarse features concurrently at superpixels, segments, and full image levels, enforcing consistency and goodness of feature induced segmentations while maximizing discrimination among image instances. Our model innovates vision transformers on three aspects. 1) We use adaptive segment tokens instead of fixed-shape patch tokens. 2) We create a token hierarchy by inserting graph pooling between transformer blocks, naturally producing consistent multi-scale segmentations while increasing the segment size and reducing the number of tokens. 3) We produce hierarchical image segmentation for free while training for recognition by maximizing image-wise discrimination. Our work delivers the first concurrent recognition and hierarchical segmentation model without any supervision. Validated on ImageNet and PASCAL VOC, it achieves better recognition and segmentation with higher computational efficiency.

Bengio and LeCun see no reason why deep learning systems cannot be made to reason. Said Bengio, "Humans also use some kind of neural nets in their brains, and I believe that there are ways to get to human-like reasoning with deep learning architectures." It was 10 years ago, in 2012, that deep learning made its breakthrough, when an innovative algorithm for classifying images based on multi-layered neural networks suddenly turned out to do spectacularly better than all algorithms before it. That breakthrough has led to deep learning's adoption in domains like speech and image recognition, automatic translation and transcription, and robotics. As deep learning was embedded into ever-more everyday applications, more and more examples of what can go wrong also surfaced: artificial intelligence (AI) systems that discriminate, confirm stereotypes, make inscrutable decisions and require a lot of data and sometimes also a huge amount of energy.

Google is proud to be a Platinum Sponsor of the European Conference on Computer Vision (ECCV 2022), a premier forum for the dissemination of research in computer vision and machine learning (ML). This year, ECCV 2022 will be held as a hybrid event, in person in Tel Aviv, Israel with virtual attendance as an option. Google has a strong presence at this year's conference with over 60 accepted publications and active involvement in a number of workshops and tutorials. We look forward to sharing some of our extensive research and expanding our partnership with the broader ML research community. We hope you'll visit our on-site or virtual booths to learn more about the research we're presenting at ECCV 2022, including several demos and opportunities to connect with our researchers.

Governments are paying growing attention to the potential of artificial intelligence – the simulation of human intelligence processes by machines – to enhance what they do. To explore how public authorities are approaching the use of AI for tasks related to public finances, Global Government Fintech – the sister title of Global Government Forum – convened an international panel on 4 October 2022 for a webinar titled'How can AI help public authorities save money and deliver better outcomes?'. The discussion, organised in partnership with SAS and Intel, highlighted how AI is already helping departments to deliver results. But also that AI remains very much an emerging and, to many, rather nebulous field with many hurdles to clear before widespread use. "Discussions of artificial intelligence often bring up connotations of an Orwellian nature, dystopian futures, Frankenstein…" said Peter Kerstens, advisor, technological innovation & cyber security at the European Commission's Financial Services Department.

ABSTRACT Currently, research in robotics, artificial intelligence and drones are advancing exponentially, they are directly or indirectly related to various areas of the economy, from agriculture to industry. With this context, this project covers these topics guiding them, seeking to provide a framework that is capable of helping to develop new future researchers. For this, we use an aerial vehicle that works autonomously and is capable of mapping the scenario and providing useful information to the end user. This occurs from a communication between a simple programming language (Scratch) and one of the most important and efficient robot operating systems today (ROS). This is how we managed to develop a tool capable of generating a 3D map and detecting objects using the camera attached to the drone. Although this tool can be used in the advanced fields of industry, it is also an important advance for the research sector. The implementation of this tool in intermediate-level institutions is aspired to provide the ability to carry out high-level projects from a simple programming language.

ABSTRACT: Commercial drones are not yet dimensioned to perform indoor autonomous tasks, since they use GPS for their location in the environment. When it comes to a space with physical obstacles (walls, metal, etc.) between the communication of the drone and the satellites that allow the precise location of the same, there is great difficulty in finding the satellites or it generates interference for this location. This problem can cause an unexpected action of the drone, a collision and a possible accident can occur, The work to follow presents the development of a drone capable of operating in a physical space (indoor), without the need for GPS. In this proposal, a prototype of a system for detecting the distance (lidar) that the drone is from the walls is also developed, with the aim of being able to take this information as the location of the drone.

Recent advances in TCP congestion control (CC) have achieved tremendous success with deep reinforcement learning (RL) approaches, which use feedforward neural networks (NN) to learn complex environment conditions and make better decisions. However, such "black-box" policies lack interpretability and reliability, and often, they need to operate outside the traditional TCP datapath due to the use of complex NNs. This paper proposes a novel two-stage solution to achieve the best of both worlds: first to train a deep RL agent, then distill its (over-)parameterized NN policy into white-box, light-weight rules in the form of symbolic expressions that are much easier to understand and to implement in constrained environments. At the core of our proposal is a novel symbolic branching algorithm that enables the rule to be aware of the context in terms of various network conditions, eventually converting the NN policy into a symbolic tree. The distilled symbolic rules preserve and often improve performance over state-of-the-art NN policies while being faster and simpler than a standard neural network. We validate the performance of our distilled symbolic rules on both simulation and emulation environments.

Memorization presents a challenge for several constrained Natural Language Generation (NLG) tasks such as Neural Machine Translation (NMT), wherein the proclivity of neural models to memorize noisy and atypical samples reacts adversely with the noisy (web crawled) datasets. However, previous studies of memorization in constrained NLG tasks have only focused on counterfactual memorization, linking it to the problem of hallucinations. In this work, we propose a new, inexpensive algorithm for extractive memorization (exact training data generation under insufficient context) in constrained sequence generation tasks and use it to study extractive memorization and its effects in NMT. We demonstrate that extractive memorization poses a serious threat to NMT reliability by qualitatively and quantitatively characterizing the memorized samples as well as the model behavior in their vicinity. Based on empirical observations, we develop a simple algorithm which elicits non-memorized translations of memorized samples from the same model, for a large fraction of such samples. Finally, we show that the proposed algorithm could also be leveraged to mitigate memorization in the model through finetuning. We have released the code to reproduce our results at https://github.com/vyraun/Finding-Memo.

Actor learning and critic learning are two components of the outstanding and mostly used Deep Deterministic Policy Gradient (DDPG) reinforcement learning method. Since actor and critic learning plays a significant role in the overall robot's learning, the performance of the DDPG approach is relatively sensitive and unstable as a result. We propose a multi-actor-critic DDPG for reliable actor-critic learning to further enhance the performance and stability of DDPG. This multi-actor-critic DDPG is then integrated with Hindsight Experience Replay (HER) to form our new deep learning framework called AACHER. AACHER uses the average value of multiple actors or critics to substitute the single actor or critic in DDPG to increase resistance in the case when one actor or critic performs poorly. Numerous independent actors and critics can also gain knowledge from the environment more broadly. We implemented our proposed AACHER on goal-based environments: AuboReach, FetchReach-v1, FetchPush-v1, FetchSlide-v1, and FetchPickAndPlace-v1. For our experiments, we used various instances of actor/critic combinations, among which A10C10 and A20C20 were the best-performing combinations. Overall results show that AACHER outperforms the traditional algorithm (DDPG+HER) in all of the actor/critic number combinations that are used for evaluation. When used on FetchPickAndPlace-v1, the performance boost for A20C20 is as high as roughly 3.8 times the success rate in DDPG+HER.