There is a growing concern about typically opaque decision-making with high-performance machine learning algorithms. Providing an explanation of the reasoning process in domain-specific terms can be crucial for adoption in risk-sensitive domains such as healthcare. We argue that machine learning algorithms should be interpretable by design and that the language in which these interpretations are expressed should be domain- and task-dependent. Consequently, we base our model's prediction on a family of user-defined and task-specific binary functions of the data, each having a clear interpretation to the end-user. We then minimize the expected number of queries needed for accurate prediction on any given input. As the solution is generally intractable, following prior work, we choose the queries sequentially based on information gain. However, in contrast to previous work, we need not assume the queries are conditionally independent. Instead, we leverage a stochastic generative model (VAE) and an MCMC algorithm (Unadjusted Langevin) to select the most informative query about the input based on previous query-answers. This enables the online determination of a query chain of whatever depth is required to resolve prediction ambiguities. Finally, experiments on vision and NLP tasks demonstrate the efficacy of our approach and its superiority over post-hoc explanations.

Artificial Intelligence is no longer just a cliched topic in the entertainment world. AI today has developed into a'must-have' for every vertical from the government to basic applications on our gadgets. The use and dependency of AI especially in businesses have increased multifold, be it for understanding the customers or developing new products to suit the needs of the users. While the use of AI is opening up never-seen-before opportunities and possibilities for organizations across verticals, it also brings in incredible responsibility to safeguard the data and ensure transparency. Ethics seem to play a bigger part in ensuring that organizations take up a responsible path in using AI for their businesses.

After recapturing Kherson city, Ukraine kept Russian forces guessing about their next move, pinning down occupying troops in defensive positions and rendering them unavailable for offensive operations. Some 30,000 Russian troops that withdrew from the west bank of the Dnieper river earlier this month were entrenching themselves in the Zaporizhia and Kherson regions during the 39th week of the war, deputy head of Ukrainian military intelligence Major-General Vadym Skibitskyi, told the Kyiv Post. "[The Russians] are waiting for our liberation offensive, that's why they have created a defensive line in Kherson, another on the administrative border of [Kherson and] Crimea, and another in the northern Crimea region," Skibitskiy said. "The enemy is on the defensive in the Zaporizhzhia direction," said Ukraine's general staff. "In the Kryvyi Rih and Kherson directions, the enemy is creating an echeloned defence system, improving fortification equipment and logistical support of advanced units, and not stopping artillery fire at the positions of our troops and settlements on the right bank of the Dnipro River."

AI is approaching the next level of maturity, coming out of the hype cycle, says Gartner. Its adoption is expanding across industries beyond automation to building new-generation intelligent products and services for business growth. However, half of them acknowledge that they don't have skilled talent to make the most of AI advances. This is where experienced AI consulting firms come in to help. The market of AI consulting is vast, ranging from tech giants like IBM and Accenture to Big 4 firms and smaller-scale innovators.

The government workscape is changing rapidly. This major shift is multifaceted with jobs switching to remote, an overall struggle to fill roles, budget cuts and the automation of many tasks. However, one thing is clear: Government agencies are being forced to do more with less. People are fleeing the public sector for private-sector jobs. In fact, the number of private-sector jobs has now surpassed its pre-pandemic level.

How would an artificial intelligence (AI) decide what to do? One common approach in AI research is called "reinforcement learning". Reinforcement learning gives the software a "reward" defined in some way, and lets the software figure out how to maximise the reward. This approach has produced some excellent results, such as building software agents that defeat humans at games like chess and Go, or creating new designs for nuclear fusion reactors. However, we might want to hold off on making reinforcement learning agents too flexible and effective.

Center for American Security's Fred Fleitz unpacks the national security risks posed by China's access to TikTok data and Chinese-made drones flying over Washington D.C. World Economic Forum founder and Chair Klaus Schwab recently sat down for an interview with a Chinese state media outlet and proclaimed that China was a "role model" for other nations. Schwab, 84, made these comments during an interview with CGTN's Tian Wei on the sidelines of last week's APEC CEO Summit in Bangkok, Thailand. Schwab said he respected China's "tremendous" achievements at modernizing its economy over the last 40 years. FILE: World Economic Forum (WEF) founder and Executive Chairman Klaus Schwab sits, as German Chancellor Olaf Scholz (not pictured) addresses the delegates, during the last day of the WEF in Davos, Switzerland May 26, 2022. "I think it's a role model for many countries," Schwab said, before qualifying that he thinks each country should make its own decisions about what system it wants to adapt.

Air transportation is essential for society, and it is increasing gradually due to its importance. To improve the airspace operation, new technologies are under development, such as Unmanned Aircraft Systems (UAS). In fact, in the past few years, there has been a growth in UAS numbers in segregated airspace. However, there is an interest in integrating these aircraft into the National Airspace System (NAS). The UAS is vital to different industries due to its advantages brought to the airspace (e.g., efficiency). Conversely, the relationship between UAS and Air Traffic Control (ATC) needs to be well-defined due to the impacts on ATC capacity these aircraft may present. Throughout the years, this impact may be lower than it is nowadays because the current lack of familiarity in this relationship contributes to higher workload levels. Thereupon, the primary goal of this research is to present a comprehensive review of the advancements in the integration of UAS in the National Airspace System (NAS) from different perspectives. We consider the challenges regarding simulation, final approach, and optimization of problems related to the interoperability of such systems in the airspace. Finally, we identify several open challenges in the field based on the existing state-of-the-art proposals.

The superior performance of supervised relation extraction (RE) methods heavily relies on a large amount of gold standard data. Recent zero-shot relation extraction methods converted the RE task to other NLP tasks and used off-the-shelf models of these NLP tasks to directly perform inference on the test data without using a large amount of RE annotation data. A potentially valuable by-product of these methods is the large-scale silver standard data. However, there is no further investigation on the use of potentially valuable silver standard data. In this paper, we propose to first detect a small amount of clean data from silver standard data and then use the selected clean data to finetune the pretrained model. We then use the finetuned model to infer relation types. We also propose a class-aware clean data detection module to consider class information when selecting clean data. The experimental results show that our method can outperform the baseline by 12% and 11% on TACRED and Wiki80 dataset in the zero-shot RE task. By using extra silver standard data of different distributions, the performance can be further improved.

Many data have an underlying dependence on spatial location; it may be weather on the Earth, a simulation on a mesh, or a registered image. Yet this feature is rarely taken advantage of, and violates common assumptions made by many neural network layers, such as translation equivariance. Further, many works that do incorporate locality fail to capture fine-grained structure. To address this, we introduce the Spatial Mixture-of-Experts (SMoE) layer, a sparsely-gated layer that learns spatial structure in the input domain and routes experts at a fine-grained level to utilize it. We also develop new techniques to train SMoEs, including a self-supervised routing loss and damping expert errors. Finally, we show strong results for SMoEs on numerous tasks, and set new state-of-the-art results for medium-range weather prediction and post-processing ensemble weather forecasts.