As machine learning models grow more complex and their applications become more high-stakes, tools for explaining model predictions have become increasingly important. This has spurred a flurry of research in model explainability and has given rise to feature attribution methods such as LIME and SHAP. Despite their widespread use, evaluating and comparing different feature attribution methods remains challenging: evaluations ideally require human studies, and empirical evaluation metrics are often data-intensive or computationally prohibitive on real-world datasets. In this work, we address this issue by releasing XAI-Bench: a suite of synthetic datasets along with a library for benchmarking feature attribution algorithms. Unlike real-world datasets, synthetic datasets allow the efficient computation of conditional expected values that are needed to evaluate ground-truth Shapley values and other metrics. The synthetic datasets we release offer a wide variety of parameters that can be configured to simulate real-world data. We demonstrate the power of our library by benchmarking popular explainability techniques across several evaluation metrics and across a variety of settings. The versatility and efficiency of our library will help researchers bring their explainability methods from development to deployment. Our code is available at https://github.com/abacusai/xai-bench.

All the sessions from Transform 2021 are available on-demand now. Low-code automation platform Appian today announced that it acquired Lana Labs, a process mining company, for an undisclosed amount. Appian says that with the addition of Lana, it'll be able to deliver actionable and continuous process optimization with people, systems, and data in the same workflow. Digital transformation and the ability to adapt quickly are critical in today's business environment. That's why a growing number of companies are adopting process mining, a family of techniques that support the analysis of operational processes of event logs, with the goal of turning event data into insights and actions.

Agriculture is a both major industry and the foundation of the economy. Artificial Intelligence (AI) techniques are widely used to solve a variety of problems and to optimize the production and operation processes in the fields of agriculture, food, and bio-system engineering. The use of artificial intelligence in the agriculture supply chain is becoming more and more important while involving Artificial Intelligence ML algorithms. The main four clusters are preproduction, production, processing, and distribution. In fact, in the preproduction, ML technologies are used, especially for the predictions of given features.

These super sized rock trucks are on their way to drop a load of Platinum rich ... [ ] rock into the crusher. These trucks can carry more than 200 tons of rock at a time, and are too large for public roads. With the total operating expenses of the top mining companies worldwide reaching USD $15 billion, efficient operational methods using AI now dubbed smart mining is rapidly advancing. McKinsey estimates that by 2035, the age of smart mining achieved through autonomous mining using data analysis and digital technologies like artificial intelligence (AI) will save between $290 billion and $390 billion annually for mineral raw materials producers. The mining industry is increasingly using artificial intelligence in innovative ways to optimize processes, enhance decision-making, derive value from data, and improve safety.

We present a novel approach for data-driven modeling of the time-domain induced polarization (IP) phenomenon using variational autoencoders (VAE). VAEs are Bayesian neural networks that aim to learn a latent statistical distribution to encode extensive data sets as lower dimension representations. We collected 1 600 319 IP decay curves in various regions of Canada, the United States and Kazakhstan, and compiled them to train a deep VAE. The proposed deep learning approach is strictly unsupervised and data-driven: it does not require manual processing or ground truth labeling of IP data. Moreover, our VAE approach avoids the pitfalls of IP parametrization with the empirical Cole-Cole and Debye decomposition models, simple power-law models, or other sophisticated mechanistic models. We demonstrate four applications of VAEs to model and process IP data: (1) representative synthetic data generation, (2) unsupervised Bayesian denoising and data uncertainty estimation, (3) quantitative evaluation of the signal-to-noise ratio, and (4) automated outlier detection. We also interpret the IP compilation's latent representation and reveal a strong correlation between its first dimension and the average chargeability of IP decays. Finally, we experiment with varying VAE latent space dimensions and demonstrate that a single real-valued scalar parameter contains sufficient information to encode our extensive IP data compilation. This new finding suggests that modeling time-domain IP data using mathematical models governed by more than one free parameter is ambiguous, whereas modeling only the average chargeability is justified. A pre-trained implementation of our model -- readily applicable to new IP data from any geolocation -- is available as open-source Python code for the applied geophysics community.

Geometric deep learning (GDL), which is based on neural network architectures that incorporate and process symmetry information, has emerged as a recent paradigm in artificial intelligence. GDL bears particular promise in molecular modeling applications, in which various molecular representations with different symmetry properties and levels of abstraction exist. This review provides a structured and harmonized overview of molecular GDL, highlighting its applications in drug discovery, chemical synthesis prediction, and quantum chemistry. Emphasis is placed on the relevance of the learned molecular features and their complementarity to well-established molecular descriptors. This review provides an overview of current challenges and opportunities, and presents a forecast of the future of GDL for molecular sciences.

After months of spaceflight, an 8-minute plunge to the surface of Mars, and weeks of exploration, NASA's Perseverance rover is beginning its primary scientific task: drilling out finger-size cores of martian rock for return to Earth. If all goes well, the first drilling sample will be collected from Jezero crater, a former lakebed, by early August. Perseverance has operated well since its February landing, and it recently tested its rock storage system, using its robotic arm to stow a sampling tube into its guts. There the empty tube was imaged and then sealed for storage. “The great news is that it all worked perfectly,” says Jennifer Trosper, Perseverance's project manager at NASA's Jet Propulsion Laboratory. “We are ready to sample.” Now, 1 kilometer south of its landing site, Perseverance has reached an array of what its operating team calls paver stones—flat, white, dust-coated rocks found throughout much of the floor of Jezero crater. Here, on what is believed to be the most ancient terrain in the crater, nearly 4 billion years old, the team will direct the rover to drill and collect its sample, targeting a rock that is average in chemistry, mineralogy, and texture. The chalk-size core will be stored in an ultraclean metallic tube, one of 38 samples the rover will eventually collect, with about 30 of those likely to be returned to Earth by later missions. It will reside in the rover's belly until it is deposited in a cache on the surface near the crater's rim a year and a half from now. Whether the paver stone landscape was deposited by the lake or formed by volcanic flows isn't clear. But if it is volcanic, it might have trapped radioactive elements that a lab on Earth could analyze to determine accurate dates for the lake's existence. The drill operators don't know what to expect because the rocks are covered with sand grains and pebbles, along with some sort of purplish coating, says Ken Farley, the mission's project scientist and a geologist at the California Institute of Technology. But before drilling into the pavers, the rover will unleash one instrument that could help answer this puzzle: an abrasion bit mounted at the end of its 2-meter-long arm. After grinding into the rock, the arm will blow compressed gas to clear away the grit, giving a clear glimpse of the underlying rock. The rover can then use its arm-mounted camera and laser and x-ray probes to probe its structure and mineralogy. “I'm pretty confident we will be able to answer this question,” Farley says. Perseverance has already spotted other tantalizing sites to explore and sample in Jezero crater. In the ancient delta to its west that is the rover's destination next year, its cameras have revealed distinctive layered deposits that show the lake was high, quiet, and stable for a long time, Farley says. Above those layers lie 1-meter-wide smooth boulders that could only have been carried by floodwater later in the lake's history. This suggests the lake could have seen distinct phases in its life, which fits with a larger picture of the planet's history in which lakes were common billions of years ago, then gave way to periodic floods after the climate cooled, Farley says. A long-lived lake might have also provided the nutrients and habitat to fuel life, says Kennda Lynch, an astrobiologist at the Lunar and Planetary Institute who is unaffiliated with the mission. “This is great. I feel more confident we chose the right place to go.” Samples and measurements from Perseverance's next target, Séítah, a region of sand dunes and ridges to its west that the car-size rover has skirted past, could test that picture. Seen from orbit to be rich in olivine, a volcanic mineral, and carbonates, which can form when olivine is exposed to water and carbon dioxide, Séítah has unexpectedly complex geology, including layered terrain that might preserve signs of past life or patterns of water flow. But the rover can't drive into it without getting stuck in the dunes, so the Perseverance team devised an incursion from above and behind to access its secrets. First, the rover's miniature Ingenuity helicopter, in its ninth flight earlier this month, scouted across Séètah in a 625-meter journey, breaking records for flight duration and speed before landing on the other side of the dunes. The helicopter photographed the intersection of Séètah with the paver unit that Perseverance is now exploring—a boundary that could reveal whether the pavers continue beneath Séitah's dunes, an important fact if a volcanic date is found. And it also scouted fractures that could hold evidence of whether ancient subsurface habitats existed in Jezero. Meanwhile, from afar, Perseverance has spied fine layering in Séítah's ridges, including a prominent 40-meter-tall plateau dubbed Kodiak that, in all likelihood, marks the delta's incursion into the lakebed. Such layers could be caused by mudstones, which smother and preserve life on Earth. But the layers could have a volcanic origin, as well—and so the rover will loop south around Séètah later this year, nudging into a flat space where it can safely sample and tease out that story. Once the Séítah campaign is done, Perseverance will backtrack all the way north to its landing site, “putting the pedal to the metal,” Trosper says. From there it will continue north then west on a safe route to the looming cliff of the main delta—and the life-trapping muds entombed within it.

Now, with mining jobs hard to find, he's cleaning up the mess the industry left behind. The 68-year-old operates a bucket loader scraping away red, rocky waste dumped years ago by failed coal mine operators in a valley in the town of Clinchco, Virginia. The $17.50 an hour before overtime he makes cleaning up massive "gob piles," as the locals call them, is less than what he earned in decades as a miner. "If this work goes away, I don't know what I would do," Mullins said. Appalachia, long the heart of the U.S. coal-mining industry, may be set for a surge in jobs like Mullins' if President Joe Biden is successful in his ambitions to transition the United States to a cleaner energy economy to fight climate change.

NIR Calibration-Model Services Spectroscopy and Chemometrics News Weekly 28, 2021 NIRS NIR Spectroscopy MachineLearning Spectrometer Spectrometric Analytical Chemistry Chemical Analysis Lab Labs Laboratories Laboratory Software IoT Sensors QA QC Testing Quality LINK This week's NIR news Weekly is sponsored by Your-Company-Name-Here – NIR-spectrometers. Check out their product page … link Get the Spectroscopy and Chemometrics News Weekly in real time on Twitter @ CalibModel and follow us. The cases of aromatic ring, C O, C N and C-Cl functionalities" LINK "Combining Vis-NIR spectroscopy and advanced statistical analysis for estimation of soil chemical properties relevant for forest road construction" LINK "Use of NIRS for the assessment of meat quality traits in open-air free-range Iberian pigs" LINK "DETECTING CONTAMINANTS IN POST-CONSUMER PLASTIC PACKAGING WASTE BY A NIR HYPERSPECTRAL IMAGING-BASED CASCADE DETECTION …" (87)80084-9 LINK Infrared Spectroscopy (IR) and Near-Infrared ...

Metaspectral, a company offering technology that derives insights from AI using ultra-high-resolution, visible-to-infrared (hyperspectral) imagery, has been awarded more than $300,000 in grant funding from the CleanBC Plastics Action Fund. The fund is funded by the BC Government and administered by Alacrity Cleantech. The CleanBC Plastics Action Fund supports B.C. businesses creating value from used plastics by including more recycled material in product manufacturing to keep plastic out of landfills. Metraspectral will use this funding for the development of computer vision, artificial intelligence, and robotics designed to sort consumer waste, increase efficiency in processing materials and improve the quality of post-consumer recycled plastic. The project is slated for completion by Dec. 31, 2021.