Distribution shifts -- where the training distribution differs from the test distribution -- can substantially degrade the accuracy of machine learning (ML) systems deployed in the wild. Despite their ubiquity, these real-world distribution shifts are under-represented in the datasets widely used in the ML community today. To address this gap, we present WILDS, a curated collection of 8 benchmark datasets that reflect a diverse range of distribution shifts which naturally arise in real-world applications, such as shifts across hospitals for tumor identification; across camera traps for wildlife monitoring; and across time and location in satellite imaging and poverty mapping. On each dataset, we show that standard training results in substantially lower out-of-distribution than in-distribution performance, and that this gap remains even with models trained by existing methods for handling distribution shifts. This underscores the need for new training methods that produce models which are more robust to the types of distribution shifts that arise in practice. To facilitate method development, we provide an open-source package that automates dataset loading, contains default model architectures and hyperparameters, and standardizes evaluations. Code and leaderboards are available at this https URL.

WILDS: A Benchmark of in-the-Wild Distribution Shifts

We present PROB, an animation and model checking tool for the B method PROB's animation facilities allow users to gain confidence in their specifications, and unlike the animator provided by the B-Toolkit, the user does not have to guess the right values for the operation arguments or choice variables PROB contains a model checker and a constraint-based checker, both of which can be used to detect various errors in B specifications We present our first experiences in using PROB on several case studies, highlighting that PROB enables users to uncover errors that are not easily discovered by existing tools

ProB: A model checker for B

It has become common to publish large (billion parameter) language models that have been trained on private datasets. This paper demonstrates that in such settings, an adversary can perform a training data extraction attack to recover individual training examples by querying the language model. 
We demonstrate our attack on GPT-2, a language model trained on scrapes of the public Internet, and are able to extract hundreds of verbatim text sequences from the model's training data. These extracted examples include (public) personally identifiable information (names, phone numbers, and email addresses), IRC conversations, code, and 128-bit UUIDs. Our attack is possible even though each of the above sequences are included in just one document in the training data. 
We comprehensively evaluate our extraction attack to understand the factors that contribute to its success. Worryingly, we find that larger models are more vulnerable than smaller models. We conclude by drawing lessons and discussing possible safeguards for training large language models.

Extracting Training Data from Large Language Models

Internet of Things (IoT) related applications have emerged as an important field for both engineers and researchers, reflecting the magnitude and impact of data-related problems to be solved in contemporary business organizations especially in cloud computing. This paper first provides a functional framework that identifies the acquisition, management, processing and mining areas of IoT big data, and several associated technical modules are defined and described in terms of their key characteristics and capabilities. Then current research in IoT application is analyzed, moreover, the challenges and opportunities associated with IoT big data research are identified. We also report a study of critical IoT application publications and research topics based on related academic and industry publications. Finally, some open issues and some typical examples are given under the proposed IoT-related research framework.

IoT-Based Big Data Storage Systems in Cloud Computing: Perspectives and Challenges

Learning visual representations of medical images is core to medical image understanding but its progress has been held back by the small size of hand-labeled datasets. Existing work commonly relies on transferring weights from ImageNet pretraining, which is suboptimal due to drastically different image characteristics, or rule-based label extraction from the textual report data paired with medical images, which is inaccurate and hard to generalize. We propose an alternative unsupervised strategy to learn medical visual representations directly from the naturally occurring pairing of images and textual data. Our method of pretraining medical image encoders with the paired text data via a bidirectional contrastive objective between the two modalities is domain-agnostic, and requires no additional expert input. We test our method by transferring our pretrained weights to 4 medical image classification tasks and 2 zero-shot retrieval tasks, and show that our method leads to image representations that considerably outperform strong baselines in most settings. Notably, in all 4 classification tasks, our method requires only 10% as much labeled training data as an ImageNet initialized counterpart to achieve better or comparable performance, demonstrating superior data efficiency.

/pdf/contrastive-learning-of-medical-visual-representations-from-15x0tzocxw.pdf

Contrastive Learning of Medical Visual Representations from Paired Images and Text

Wireless sensor network (WSN) has now become a newfangled area of research. WSN is an interconnection of sensor nodes which are installed in a region for a particular task. Sensors sense their surrounding physical attributes and transmit the acquired information to the dedicated node or base station. There are some challenges associated with WSN: Energy efficiency, security, coverage, connectivity, responsiveness, etc. WSN has large number of application areas. Energy efficiency of the network is a major concern for some applications because battery replacement is not easy. Clustering is used to enhance energy efficiency of the network. When sensor nodes transmit information, some amount of energy dissipated. But when nodes are not in transmitting state, still some amount of energy is consumed in listening to data. In this paper, we proposed a hybrid of probabilistic and Sleep–wakeup clusterings to decrease the energy consumption due to idle listening. The proposed method adds the energy efficiency by using adaptive probabilistic approach with a sleep–wakeup schedule. Results are simulated in MATLAB. The analysis of results reveals that the duration of first-node dead and last-node dead while considering round number is increased in comparison of LEACH, which ultimately shows increased energy efficiency of the system.

A Hybrid of Scheduling and Probabilistic Approach to Decrease the Effect of Idle Listening in WSN

This is the era of modern and highly sophisticated research which is on-going at various geographical locations and there are so many consequences that are also associated with the security and ethical responsibility. Hence knowingly or unknowingly the human civilization would face the kind of traumas and pandemics which has already been happened in the recent past. This chapter discusses these kind of consequences and their remedies with the latest technological developments whether it is the security, tracking, IOT based healthcare systems, healthcare seminars, conferences, etc. The healthcare systems touches every acumen of the human life and impact greatly the day-to-day human life. The tracking, surveillance and monitoring, etc. are not that difficult due to the recent advancements and IOT based researches. The robotic diagnostic arm system, robotic patrolling and surveillance, and cobotics are the ideal inventions of the modern era by which the medical practitioners now easily can monitor, diagnose as well as do the surgery to the remotely placed patient or the medication seekers sitting at geographically distant places. Technological interventions helped human life easier, but at the same time when the minute level of mistakes will create the havocs and its capable of and can destroy the entire human race in very less time just like Covid’19 is the latest example. These pandemics are very difficult to control, hence, there is an utmost care needed, while developing and performing the research with these viruses whether to control, test, vaccine researches, trials on these vaccines, etc. The pandemics and their threat to the humans and any living being are dangerous in any form. The entire world shown the united support and courage to control the Covid’19 as the latest pandemic, but it is very necessary that people understand the necessity to act wisely, this chapter also discusses if the situations wrong and how to control the lockdown situations using sophisticated surveillance using drone cameras, IOT devices, robotics, unmanned vehicles, etc. used to control stampede and festive gathering or any other kind of people gatherings. Advanced people monitoring and machine learning based predictions systems utilized in order to reduce such situations where the gathering control is difficult. These devices helping the government and the police cops to perform the duties and to avoid any such situations where this is uncontrollable. The world economy and mental health aspects, lockdown control and employee sanitization automation systems are also discussed in this chapter.

Emerging Technologies for Pandemic and Its Impact

This paper proposes a static index pruning method for phrase queries which is based on the cohesive similarity between terms. The co-occurrence between terms is model by term's cohesiveness within document. The less relevant terms gets pruned away while assuring that there is no change in the top-k results. The proposed method creates an effective pruned index. This method also considers the term proximity based on the term frequency and the terms informative ness. The experiments were conducted on a number of different standard text collections, and analysis of the results shows promising results and is comparable with the existing static pruning method. It is an advantage of the proposed approach that it can be applies to standard inverted index for phrase queries also.

http://ieeexplore.ieee.org/iel7/7058462/7065338/07065545.pdf

Phrase Term Static Index Pruning Based on the Term Cohesiveness

In recent times, Covid-19 has changed the dimensions of the educational industry. Universities across the global are focusing on the changing trends, technologies, and practices influencing the teaching and learning among teachers and students. This research paper mainly focuses on the emerging technologies in the Covid-19 providing about the real-time examples and insight the brief about the transformational shift how the universities are architect the various ecosystems both for instructors and learners. The relevant dataset of exam, quizzes, etc., from heterogeneous department were utilized for proposed methodology. The research work also includes the implications and challenges faced by the universities while implementing these technologies. The accuracy obtained was higher in the twenty-first century e-learning tools and lesser in all other cases as well as for the legacy system. The performance was observed, and various inferences were discussed with the effective delivery of the teaching material and their issues.

Performance Comparison for E-Learning and Tools in Twenty-First Century with Legacy System Using Classification Approach

The pursuit of happiness is the ultimate purpose of human life in modern civilization. Happiness is defined in Indian culture as a mood or quality that occurs at birth in a person and distinguishes it from other species. Meditation, as the phrase from the Ashtavakra Gita suggests, is a spiritual instrument for increasing happiness in life. Meditation is a term that refers to a set of techniques that are often used to relieve stress and increase overall well-being. These days, HRV (heart rate variability) appears to be on each person mind, at least among those who want to track their own or their customers’ sleep, recuperation, performance, or inclusive mental health. The findings of this study are both inspiring and useful, as the current analysis clearly reveals that meditation has a considerable impact on heart rate dynamics. The results of heart rate variability metrics provide a quantifiable measurement of this. It plainly states that meditation improves one’s health, which leads to happiness as the ultimate objective. In this research article the ECG based health analysis and happiness measurement demonstrated which help oneself to get a maximum amount of happiness through the use of of yoga and natural resources.

Akhilesh Kumar Sharma

Papers

A Hybrid of Scheduling and Probabilistic Approach to Decrease the Effect of Idle Listening in WSN

Emerging Technologies for Pandemic and Its Impact

Phrase Term Static Index Pruning Based on the Term Cohesiveness

Performance Comparison for E-Learning and Tools in Twenty-First Century with Legacy System Using Classification Approach

Meditation for Happy and Healthy Life: A Quantitative Assessment using Heart Rate Variability using ECG