These opportunities require a significant shift in the way the BPMS operates and interacts with its operators(both human and digital agents). While traditional BPMSs encode pre-defined flows and rules, an ABPMS is able to reason about the current state of the process(or across several processes) to determine a course of action that improves the performance of the process. To fully exploit this capability, the ABPMS needs a degree of autonomy. Naturally, this autonomy needs to be framed by operational assumptions, goals, and environmental constraints. Also, ABPMSs need to engage conversationally with human agents, they need to explain their actions, and they need to recommend adaptations or improvements in the way the process is performed. This manifesto outlined a number of research challenges that need to be overcome to realize systems that exhibit these characteristics.

Active learning is commonly used to train label-efficient models by adaptively selecting the most informative queries. However, most active learning strategies are designed to either learn a representation of the data (e.g., embedding or metric learning) or perform well on a task (e.g., classification) on the data. However, many machine learning tasks involve a combination of both representation learning and a task-specific goal. Motivated by this, we propose a novel unified query framework that can be applied to any problem in which a key component is learning a representation of the data that reflects similarity. Our approach builds on similarity or nearest neighbor (NN) queries which seek to select samples that result in improved embeddings. The queries consist of a reference and a set of objects, with an oracle selecting the object most similar (i.e., nearest) to the reference. In order to reduce the number of solicited queries, they are chosen adaptively according to an information theoretic criterion. We demonstrate the effectiveness of the proposed strategy on two tasks -- active metric learning and active classification -- using a variety of synthetic and real world datasets. In particular, we demonstrate that actively selected NN queries outperform recently developed active triplet selection methods in a deep metric learning setting. Further, we show that in classification, actively selecting class labels can be reformulated as a process of selecting the most informative NN query, allowing direct application of our method.

Variational approximations to Gaussian processes (GPs) typically use a small set of inducing points to form a low-rank approximation to the covariance matrix. In this work, we instead exploit a sparse approximation of the precision matrix. We propose variational nearest neighbor Gaussian process (VNNGP), which introduces a prior that only retains correlations within K nearest-neighboring observations, thereby inducing sparse precision structure. Using the variational framework, VNNGP's objective can be factorized over both observations and inducing points, enabling stochastic optimization with a time complexity of O($K^3$). Hence, we can arbitrarily scale the inducing point size, even to the point of putting inducing points at every observed location. We compare VNNGP to other scalable GPs through various experiments, and demonstrate that VNNGP (1) can dramatically outperform low-rank methods, and (2) is less prone to overfitting than other nearest neighbor methods.

Writing about A.I. has given me the opportunity to think more about datascience, programming and technology trends for new professionals. Picture this, a young woman in Africa starts her her own startup to meet unmet needs, the year is 2030. Will she need to even hire developers or can her small IT team handle the engineering of the product? Are we heading to a world where the software revolution will truly be democratized? This is an Op-ed about an emerging and somewhat speculative field that is showing real signs of potential.

This story was originally published by the Guardian and is reproduced here as part of the Climate Desk collaboration. There are an estimated 73,300 species of tree on Earth, 9,000 of which have yet to be discovered, according to a global count of tree species by thousands of researchers who used second world war codebreaking techniques created at Bletchley Park to evaluate the number of unknown species. Researchers working on the ground in 90 countries collected information on 38 million trees, sometimes walking for days and camping in remote places to reach them. The study found there are about 14 percent more tree species than previously reported and that a third of undiscovered tree species are rare, meaning they could be vulnerable to extinction by human-driven changes in land use and the climate crisis. "It is a massive effort for the whole world to document our forests," said Jingjing Liang, a lead author of the paper and professor of quantitative forest ecology at Purdue University in Indiana, US. "Counting the number of tree species worldwide is like a puzzle with pieces spreading all over the world. We solved it together as a team, each sharing our own piece."

More and more people have started to pay attention to artificial intelligence (AI) in recent years. According to Edelman's special report on tech within its annual Trust Barometer report, people around the world have shown concern that AI and robots could replace human workers. As a result, fewer people are willing to share their personal data, as their trust in the media, online social platforms and search engines seems to have declined. Some say the chasm between trust and technology has formed for good reasons: For most of AI's existence, there hasn't been much regulation around it. At times, the rules may seem a bit loose and opaque for just how world-changing it could be.

The Meantime To Failure (MTTF) is a statistic used for system analysis in several knowledge areas. This value represents the average time to the system enters into one of the possible states of fault, without considering system repairs. Although MTTF be considered to analyze systems with fault states, it also can be used to perform analysis on processes, since it can be used to represent the meantime to one process finishes, given that, processes can be represented by state machine models. This work presents a method to compute MTTF from Continuous Time Markov Chain (CTMC) models. There are no arguments that demonstrate that this method performs better than other methods, but this method has a simpler implementation and is intuitive. This method also allows computing the absorption probabilities and the average holding time of each state without additional steps.

Fairness and robustness are often considered as orthogonal dimensions when evaluating machine learning models. However, recent work has revealed interactions between fairness and robustness, showing that fairness properties are not necessarily maintained under distribution shift. In healthcare settings, this can result in e.g. a model that performs fairly according to a selected metric in "hospital A" showing unfairness when deployed in "hospital B". While a nascent field has emerged to develop provable fair and robust models, it typically relies on strong assumptions about the shift, limiting its impact for real-world applications. In this work, we explore the settings in which recently proposed mitigation strategies are applicable by referring to a causal framing. Using examples of predictive models in dermatology and electronic health records, we show that real-world applications are complex and often invalidate the assumptions of such methods. Our work hence highlights technical, practical, and engineering gaps that prevent the development of robustly fair machine learning models for real-world applications. Finally, we discuss potential remedies at each step of the machine learning pipeline.

The digital world has been entirely transformed with the help of technological breakthroughs, and IoT (Internet of Things) is to be credited among AI (Artificial Intelligence), ML (Machine Learning), Data Science, and more. Internet of Things has been the futuristic concept of connecting and controlling our devices and items remotely. This future idea alone has brought drastic change within many industries that have seen improved processes, increased productivity, and many other benefits. However, one of the most significant contributions that IoT has made in assisting users with disabilities. We'll get to that thought in this article.

A major bottleneck in training robust Human-Activity Recognition models (HAR) is the need for large-scale labeled sensor datasets. Because labeling large amounts of sensor data is an expensive task, unsupervised and semi-supervised learning techniques have emerged that can learn good features from the data without requiring any labels. In this paper, we extend this line of research and present a novel technique called Collaborative Self-Supervised Learning (ColloSSL) which leverages unlabeled data collected from multiple devices worn by a user to learn high-quality features of the data. A key insight that underpins the design of ColloSSL is that unlabeled sensor datasets simultaneously captured by multiple devices can be viewed as natural transformations of each other, and leveraged to generate a supervisory signal for representation learning. We present three technical innovations to extend conventional self-supervised learning algorithms to a multi-device setting: a Device Selection approach which selects positive and negative devices to enable contrastive learning, a Contrastive Sampling algorithm which samples positive and negative examples in a multi-device setting, and a loss function called Multi-view Contrastive Loss which extends standard contrastive loss to a multi-device setting. Our experimental results on three multi-device datasets show that ColloSSL outperforms both fully-supervised and semi-supervised learning techniques in majority of the experiment settings, resulting in an absolute increase of upto 7.9% in F_1 score compared to the best performing baselines. We also show that ColloSSL outperforms the fully-supervised methods in a low-data regime, by just using one-tenth of the available labeled data in the best case.