Robert Bosch (India)

Deep learning methods have become the de-facto standard for challenging image processing tasks such as image classification. One major hurdle of deep learning approaches is that large sets of labeled data are necessary, which can be prohibitively costly to obtain, particularly in medical image diagnosis applications. Active learning techniques can alleviate this labeling effort. In this paper we investigate some recently proposed methods for active learning with high-dimensional data and convolutional neural network classifiers. We compare ensemble-based methods against Monte-Carlo Dropout and geometric approaches. We find that ensembles perform better and lead to more calibrated predictive uncertainties, which are the basis for many active learning algorithms. To investigate why Monte-Carlo Dropout uncertainties perform worse, we explore potential differences in isolation in a series of experiments. We show results for MNIST and CIFAR-10, on which we achieve a test set accuracy of 90% with roughly 12,200 labeled images, and initial results on ImageNet. Additionally, we show results on a large, highly class-imbalanced diabetic retinopathy dataset. We observe that the ensemble-based active learning effectively counteracts this imbalance during acquisition.

3D imaging technologies are applied in numerous areas, including self-driving cars, drones, and robots, and in advanced industrial, medical, scientific, and consumer applications. 3D imaging is usually accomplished by finding the distance to multiple points on an object or in a scene, and then creating a point cloud of those range measurements. Different methods can be used for the ranging. Some of these methods, such as stereovision, rely on processing 2D images. Other techniques estimate the distance more directly by measuring the round-trip delay of an ultrasonic or electromagnetic wave to the object. Ultrasonic waves suffer large losses in air and cannot reach distances beyond a few meters. Radars and lidars use electromagnetic waves in radio and optical spectra, respectively. The shorter wavelengths of the optical waves compared to the radio frequency waves translates into better resolution, and a more favorable choice for 3D imaging. The integration of lidars on electronic and photonic chips can lower their cost, size, and power consumption, making them affordable and accessible to all the abovementioned applications. This review article explains different lidar aspects and design choices, such as optical modulation and detection techniques, and point cloud generation by means of beam-steering or flashing an entire scene. Popular lidar architectures and circuits are presented, and the superiority of the FMCW lidar is discussed in terms of range resolution, receiver sensitivity, and compatibility with emerging technologies. At the end, an electronic-photonic integrated circuit for a micro-imaging FMCW lidar is presented as an example.

The book Statistical Rethinking explores techniques taught in a graduate course on statistics with more focus on understanding and relating with practical problems from a Bayesian perspective. The ...

In this paper, we introduce NLP resources for 11 major Indian languages from two major language families. These resources include: (a) large-scale sentence-level monolingual corpora, (b) pre-trained word embeddings, (c) pre-trained language models, and (d) multiple NLU evaluation datasets (IndicGLUE benchmark). The monolingual corpora contains a total of 8.8 billion tokens across all 11 languages and Indian English, primarily sourced from news crawls. The word embeddings are based on FastText, hence suitable for handling morphological complexity of Indian languages. The pre-trained language models are based on the compact ALBERT model. Lastly, we compile the IndicGLUE benchmark for Indian language NLU. To this end, we create datasets for the following tasks: Article Genre Classification, Headline Prediction, Wikipedia Section-Title Prediction, Cloze-style Multiple choice QA, Winograd NLI and COPA. We also include publicly available datasets for some Indic languages for tasks like Named Entity Recognition, Cross-lingual Sentence Retrieval, Paraphrase detection, etc. Our embeddings are competitive or better than existing pre-trained embeddings on multiple tasks. We hope that the availability of the dataset will accelerate Indic NLP research which has the potential to impact more than a billion people. It can also help the community in evaluating advances in NLP over a more diverse pool of languages. The data and models are available at https: //indicnlp.ai4bharat.org.

In response to recent criticism of gradient-based visualization techniques, we propose a new methodology to generate visual explanations for deep Convolutional Neural Networks (CNN) - based models. Our approach - Ablation-based Class Activation Mapping (Ablation CAM) uses ablation analysis to determine the importance (weights) of individual feature map units w.r.t. class. Further, this is used to produce a coarse localization map highlighting the important regions in the image for predicting the concept. Our objective and subjective evaluations show that this gradient-free approach works better than state-of-the-art Grad-CAM technique. Moreover, further experiments are carried out to show that Ablation-CAM is class discriminative as well as can be used to evaluate trust in a model.

The metabolome of an organism depends on environmental factors and intracellular regulation and provides information about the physiological conditions. Metabolomics helps to understand disease progression in clinical settings or estimate metabolite overproduction for metabolic engineering. The most popular analytical metabolomics platform is mass spectrometry (MS). However, MS metabolome data analysis is complicated, since metabolites interact nonlinearly, and the data structures themselves are complex. Machine learning methods have become immensely popular for statistical analysis due to the inherent nonlinear data representation and the ability to process large and heterogeneous data rapidly. In this review, we address recent developments in using machine learning for processing MS spectra and show how machine learning generates new biological insights. In particular, supervised machine learning has great potential in metabolomics research because of the ability to supply quantitative predictions. We review here commonly used tools, such as random forest, support vector machines, artificial neural networks, and genetic algorithms. During processing steps, the supervised machine learning methods help peak picking, normalization, and missing data imputation. For knowledge-driven analysis, machine learning contributes to biomarker detection, classification and regression, biochemical pathway identification, and carbon flux determination. Of important relevance is the combination of different omics data to identify the contributions of the various regulatory levels. Our overview of the recent publications also highlights that data quality determines analysis quality, but also adds to the challenge of choosing the right model for the data. Machine learning methods applied to MS-based metabolomics ease data analysis and can support clinical decisions, guide metabolic engineering, and stimulate fundamental biological discoveries.

Learning acoustic models directly from the raw waveform data with minimal processing is challenging. Current waveform-based models have generally used very few (~2) convolutional layers, which might be insufficient for building high-level discriminative features. In this work, we propose very deep convolutional neural networks (CNNs) that directly use time-domain waveforms as inputs. Our CNNs, with up to 34 weight layers, are efficient to optimize over very long sequences (e.g., vector of size 32000), necessary for processing acoustic waveforms. This is achieved through batch normalization, residual learning, and a careful design of down-sampling in the initial layers. Our networks are fully convolutional, without the use of fully connected layers and dropout, to maximize representation learning. We use a large receptive field in the first convolutional layer to mimic bandpass filters, but very small receptive fields subsequently to control the model capacity. We demonstrate the performance gains with the deeper models. Our evaluation shows that the CNN with 18 weight layers outperforms the CNN with 3 weight layers by over 15% in absolute accuracy for an environmental sound recognition task and is competitive with the performance of models using log-mel features.

Systems biology has experienced dramatic growth in the number, size, and complexity of computational models. To reproduce simulation results and reuse models, researchers must exchange unambiguous model descriptions. We review the latest edition of the Systems Biology Markup Language (SBML), a format designed for this purpose. A community of modelers and software authors developed SBML Level 3 over the past decade. Its modular form consists of a core suited to representing reaction-based models and packages that extend the core with features suited to other model types including constraint-based models, reaction-diffusion models, logical network models, and rule-based models. The format leverages two decades of SBML and a rich software ecosystem that transformed how systems biologists build and interact with models. More recently, the rise of multiscale models of whole cells and organs, and new data sources such as single-cell measurements and live imaging, has precipitated new ways of integrating data with models. We provide our perspectives on the challenges presented by these developments and how SBML Level 3 provides the foundation needed to support this evolution.

Today, the Internet of Things (IoT) comprises vertically oriented platforms for things. Developers who want to use them need to negotiate access individually and adapt to the platform-specific API and information models. Having to perform these actions for each platform often outweighs the possible gains from adapting applications to multiple platforms. This fragmentation of the IoT and the missing interoperability result in high entry barriers for developers and prevent the emergence of broadly accepted IoT ecosystems. The BIG IoT (Bridging the Interoperability Gap of the IoT) project aims to ignite an IoT ecosystem as part of the European Platforms Initiative. As part of the project, researchers have devised an IoT ecosystem architecture. It employs five interoperability patterns that enable cross-platform interoperability and can help establish successful IoT ecosystems.

Deep learning based 3D reconstruction techniques have recently achieved impressive results. However, while state-of-the-art methods are able to output complex 3D geometry, it is not clear how to extend these results to time-varying topologies. Approaches treating each time step individually lack continuity and exhibit slow inference, while traditional 4D reconstruction methods often utilize a template model or discretize the 4D space at fixed resolution. In this work, we present Occupancy Flow, a novel spatio-temporal representation of time-varying 3D geometry with implicit correspondences. Towards this goal, we learn a temporally and spatially continuous vector field which assigns a motion vector to every point in space and time. In order to perform dense 4D reconstruction from images or sparse point clouds, we combine our method with a continuous 3D representation. Implicitly, our model yields correspondences over time, thus enabling fast inference while providing a sound physical description of the temporal dynamics. We show that our method can be used for interpolation and reconstruction tasks, and demonstrate the accuracy of the learned correspondences. We believe that Occupancy Flow is a promising new 4D representation which will be useful for a variety of spatio-temporal reconstruction tasks.

The resource‐based view () is one of the most influential perspectives in the organizational sciences. Although entrepreneurship researchers are increasingly leveraging the tenets, it emerged in strategic management. Despite some important similarities between entrepreneurship and strategic management, there are also important differences, raising questions as to whether and to what extent the needs to be adapted for the entrepreneurship field. As a first step toward answering these questions, this study focuses on resources as the fundamental building block of the and presents a content‐analytical comparison of researchers' and practicing entrepreneurs' resource conceptualizations to derive similarities and differences between established theory and entrepreneurial practice. We find that although the two conceptualizations exhibit some overlap, there are also important differences in the emphasis on different dimensions of resources and ownership requirements, as well as in the understanding of how those resources shape outcomes. These results suggest important contextual conditions when applying the RBV's tenets within the field of entrepreneurship.

The Next Steps in Signaling (NSIS) working group is considering protocols for signaling information about a data flow along its path in the network. The NSIS suite of protocols is envisioned to support various signaling applications that need to install and/or manipulate such state in the network. Based on existing work on signaling requirements, this document proposes an architectural framework for these signaling protocols. This document provides a model for the network entities that take part in such signaling, and for the relationship between signaling and the rest of network operation. We decompose the overall signaling protocol suite into a generic (lower) layer, with separate upper layers for each specific signaling application.

Significant changes in society were emphasized as being required to achieve Sustainable Development Goals, a need which was further intensified with the emergence of the pandemic. The prospective society should be directed towards sustainable development, a process in which technology plays a crucial role. The proposed study discusses the technological potential for attaining the Sustainable Development Goals via disruptive technologies. This study further analyzes the outcome of disruptive technologies from the aspects of product development, health care transformation, a pandemic case study, nature-inclusive business models, smart cities and villages. These outcomes are mapped as a direct influence on Sustainable Development Goals 3, 8, 9 and 11. Various disruptive technologies and the ways in which the Sustainable Development Goals are influenced are elaborated. The investigation into the potential of disruptive technologies highlighted that Industry 5.0 and Society 5.0 are the most supportive development to underpin the efforts to achieve the Sustainable Development Goals. The study proposes the scenario where both Industry 5.0 and Society 5.0 are integrated to form smart cities and villages where the prospects of achieving Sustainable Development Goals are more favorable due to the integrated framework and Sustainable Development Goals’ interactions. Furthermore, the study proposes an integrated framework for including new age technologies to establish the concepts of Industry 5.0 and Society 5.0 integrated into smart cities and villages. The corresponding influence on the Sustainable Development Goals are also mapped. A SWOT analysis is performed to assess the proposed integrated approach to achieve Sustainable Development Goals. Ultimately, this study can assist the industrialist, policy makers and researchers in envisioning Sustainable Development Goals from technological perspectives.

Arbiter Physically Unclonable Functions (APUFs), while being relatively lightweight, are extremely vulnerable to modeling attacks. Hence, various compositions of APUFs such as XOR APUF and Lightweight Secure PUF have been proposed to be secure alternatives. Previous research has demonstrated that PUF compositions have two major challenges to overcome: vulnerability against modeling and statistical attacks, and lack of reliability. In this paper, we introduce a multiplexer-based composition of APUFs, denoted as MPUF, to simultaneously overcome these challenges. In addition to the basic MPUF design, we propose two MPUF variants namely cMPUF and rMPUF to improve the robustness against cryptanalysis and reliability-based modeling attack, respectively. An rMPUF demonstrates enhanced robustness against the reliability-based modeling attack, while even the well-known XOR APUF, otherwise robust to machine learning based modeling attacks, has been modeled using the same technique with linear data and time complexities. The rMPUF can provide a good trade-off between security and hardware overhead while maintaining a significantly higher reliability level than any practical XOR APUF instance. Moreover, MPUF variants are the first APUF compositions, to the best of our knowledge, that can achieve Strict Avalanche Criterion without using any additional input network (or hardware) for challenge transformation. Finally, we validate our theoretical findings using Matlab-based simulations of MPUFs.

The Progetto Lombardo Atero-Trombosi (PLAT) Study was a prospective, multicenter, multidisciplinary study of the association among hemostatic variables, conventional risk factors, and atherothrombotic events in four groups of patients with preexisting vascular ischemic disease (335 myocardial infarction survivors, 123 patients with stable angina pectoris, 160 with transient ischemic attacks, and 335 with peripheral vascular disease). In the myocardial infarction group, univariate analysis showed that atherothrombotic events were associated with high fibrinogen (p = 0.001), factor VIII:C (p less than 0.001), and von Willebrand factor antigen (vWF:Ag) (p = 0.004) levels and with low high density lipoprotein cholesterol (p = 0.043), factor VII (p = 0.019), and protein C (p = 0.044) levels; multivariate analysis produced associations with high fibrinogen and factor VIII:C levels and low protein C levels. By both univariate and multivariate analysis, events in the angina pectoris group were associated with high vWF:Ag (p = 0.026) and leukocyte (p = 0.033) levels and the presence of carotid arterial stenosis (p = 0.063); associations with high leukocyte (p = 0.037) and factor VIII:C (p = 0.186) levels, family history (p = 0.031), and diabetes (p = 0.061) were also found in the group with transient ischemic attacks. In those with peripheral vascular disease, events were associated with Fontaine stage greater than or equal to IIB (p = 0.024), high factor VIII:C levels (p = 0.073), and low protein C (p = 0.028), fibrinogen (p = 0.030), antithrombin III (p = 0.054), and factor VII (p = 0.057) levels by univariate analysis and with Fontaine stage and low fibrinogen levels by multivariate analysis.(ABSTRACT TRUNCATED AT 250 WORDS)

OBJECTIVE: A subset of myasthenia gravis patients that are seronegative for anti-acetylcholine receptor (anti-AChR) antibodies are instead seropositive for antibodies against the muscle-specific kinase (anti-MuSK-positive). Here, we test whether transfer of IgG from anti-MuSK-positive patients to mice confers impairment of the neuromuscular junction and muscle weakness. METHODS: IgG from anti-MuSK-positive myasthenia gravis patients or control IgG (seronegative for AChR and MuSK) was injected intraperitoneally (45 mg daily for 14 days) into 6-week-old female FVB/NJ and C57BL/6J mice. Changes at neuromuscular junctions in the tibialis anterior and diaphragm muscles were assessed by confocal fluorescent imaging of AChRs stained with fluorescent-alpha-bungarotoxin. Loss of function was assessed by electromyography. RESULTS: In experimental mice injected with anti-MuSK-positive patient IgG, postsynaptic AChR staining was reduced to as little as 22% of that seen in control mice. Experimental mice showed reduced apposition of the nerve terminal (labeled with antibodies against synaptophysin and neurofilament) and the postsynaptic AChR cluster (labeled with fluorescent-alpha-bungarotoxin). Mice injected with IgG from two of three anti-MuSK-positive patients lost weight and developed muscle weakness associated with a decremental electromyographic trace on repetitive nerve stimulation. INTERPRETATION: IgG from anti-MuSK-positive patients can cause myasthenia gravis when injected into mice. This may be explained by a progressive reduction in the density of postsynaptic AChR combined with changes in the nerve terminal and its relation to the postsynaptic structure.

Existing dialog datasets contain a sequence of utterances and responses without any explicit background knowledge associated with them. This has resulted in the development of models which treat conversation as a sequenceto-sequence generation task (i.e., given a sequence of utterances generate the response sequence). This is not only an overly simplistic view of conversation but it is also emphatically different from the way humans converse by heavily relying on their background knowledge about the topic (as opposed to simply relying on the previous sequence of utterances). For example, it is common for humans to (involuntarily) produce utterances which are copied or suitably modified from background articles they have read about the topic. To facilitate the development of such natural conversation models which mimic the human process of conversing, we create a new dataset containing movie chats wherein each response is explicitly generated by copying and/or modifying sentences from unstructured background knowledge such as plots, comments and reviews about the movie. We establish baseline results on this dataset (90K utterances from 9K conversations) using three different models: (i) pure generation based models which ignore the background knowledge (ii) generation based models which learn to copy information from the background knowledge when required and (iii) span prediction based models which predict the appropriate response span in the background knowledge.

Solving large complex partial differential equations (PDEs), such as those that arise in computational fluid dynamics (CFD), is a computationally expensive process. This has motivated the use of deep learning approaches to approximate the PDE solutions, yet the simulation results predicted from these approaches typically do not generalize well to truly novel scenarios. In this work, we develop a hybrid (graph) neural network that combines a traditional graph convolutional network with an embedded differentiable fluid dynamics simulator inside the network itself. By combining an actual CFD simulator (run on a much coarser resolution representation of the problem) with the graph network, we show that we can both generalize well to new situations and benefit from the substantial speedup of neural network CFD predictions, while also substantially outperforming the coarse CFD simulation alone.

The design of a silicon Strong Physical Unclonable Function (PUF) that is lightweight and stable, and which possesses a rigorous security argument, has been a fundamental problem in PUF research since its very beginnings in 2002. Various effective PUF modeling attacks, for example at CCS 2010 and CHES 2015, have shown that currently, no existing silicon PUF design can meet these requirements. In this paper, we introduce the novel Interpose PUF (iPUF) design, and rigorously prove its security against all known machine learning (ML) attacks, including any currently known reliability-based strategies that exploit the stability of single CRPs (we are the first to provide a detailed analysis of when the reliability based CMA-ES attack is successful and when it is not applicable). Furthermore, we provide simulations and confirm these in experiments with FPGA implementations of the iPUF, demonstrating its practicality. Our new iPUF architecture so solves the currently open problem of constructing practical, silicon Strong PUFs that are secure against state-of-the-art ML attacks.

Aspect extraction is a key task of fine-grained opinion mining. Although it has been studied by many researchers, it remains to be highly challenging. This paper proposes a novel unsupervised approach to make a major improvement. The approach is based on the framework of lifelong learning and is implemented with two forms of recommendations that are based on semantic similarity and aspect associations respectively. Experimental results using eight review datasets show the effectiveness of the proposed approach.