During its first keynote at Google I/O 2022, Google detailed its latest language model, LaMDA 2, and an app called AI Test Kitchen.
Category: robotics/AI – Page 1272
For instance, when training a gestational age clock model from placental methylation, a sample can only be collected after delivery of the baby and the placenta. So most samples have a gestational age greater than 30 weeks, which corresponds to moderate preterm and full-term births. For samples with a further younger gestational age, they are scarce, which makes the sample distribution seriously biased to large gestational ages and impairs the ability of the trained model to predict small ones. However, differences in gestational age as small as one week can significantly influence neonatal morbidity and mortality and long-term outcomes [18 – 23]. Hence, the model’s accuracy across the whole gestational age range becomes essential.
To solve this problem, we developed the R package eClock (ensemble-based clock). It improves the traditional machine learning strategy in handling the imbalance problem of category data [24], and combines bagging and SMOTE (Synthetic Minority Over-sampling Technique) methods to adjust the biased age distribution and predict DNAm age with an ensemble model. This is the first time applying these techniques to the clock model, bringing a new framework for clock model construction. eClock also provides other functions, such as training the traditional clock model, displaying features, and converting methylation probe/gene/DMR (DNA methylation region) values. To test the performance of the package, we used 3 different datasets, and the results show that the package can effectively improve the clock model performance on rare samples.
After launching our warehouse robot Stretch at #MODEX2022, we heard from a lot of warehouse and materials handling professionals. Check out our biggest takeaway… See more.
Launching Stretch, our autonomous warehouse robot, at MODEX 2022, we found consistent themes about the state of warehouse automation and material handling.
Summary
The human brain has the, remarkable ability to learn patterns from small amounts of data and then recognize novel instances of those patterns despite distortion and noise. Although advances in machine learning algorithms have been weakly informed by the brain since the 1940’s, they do not yet rival human performance.
On account of the improvement the Internet of things (IoTs) and smart devices, our lives have been noticeably facilitated in the past few years. Machines and devices are becoming more ingenious with the help of artificial intelligence and various sensors1,2. So, integrated circuits are necessary to provide convenient and effectual communication3 Since the first report on TENG by Wang’s group in 20124, triboelectric systems have been recognized as a proper choice to harvest and convert the energy from the environment5,6. Photodetectors, as one of the most significant types of sensors that can precisely convert incident light into electrical signals have attracted increasing attention in recent years. Various applications including photo-sensors, spectral analysis7,8, environment monitoring9, communication devices10, imaging11, take advantage of narrow band or broad band photodetectors from ultraviolet to terahertz wavelenght. Literature reviews show that the heterojunction/heterostructure based on 2D/3D materials have been widely used in PD applications. In fact, to attain high performance of PDs based heterojunction, the built-in electrical field is needed to suppress the photogenerated recombination and stimulating collection12. Although, Si based PDs offer reliably high performance results, their complexity and expensive manufacturing process have limited their expansion and adoptability for industrial purposes13,14,15. Hence, most available PDs are designed based on external power supplies such as electrochemical batteries for signal production and processing, their design not only increases the sensor’s dimension and weight, but also creates limitations for sensor maintenances16 which is not proper in the IoTs. In 2014, ZH Lin et al. and Zheng et al. represented an investigation on the self-powered PD based on TENG system3,17, and since then, self-driven PDs have been extensively investigated2,5,9,18,19,20. These devices can find potential applications in health monitoring systems such as heart checking21 and health protection from some detrimental radiation such as high levels of UV radiance22.
But in the other hand, even though TENGs could be promise for using in wearable electronics, they still inevitably have limitations in power generation, sensing range, sensitivity, and also the sensing domain for the intrinsic limitations of electrification23,24,25. Moreover, due to high voltage, low current, and alternating current output of the TENGs, they cannot be used in order to supply power to electronic devices effectively without using power management circuits (PMCs) based on the LC modules. There are several reports that describe the importance of the impedance matching of the TENG and PMC units for better energy storage efficiency of the pulsed-TENG26,27. Without using the PMC unit, there are some challenges as a result of synching the TENG, as the power supply, and the consumption element such as the PD device. These challenges include the process of matching the resistance of the device and the impedance of the TENG to achieve effective performance of the self-powered system6,28.
In this study an efficient battery-free photodetector based on bulk heterojunction SnS2 nanosheets and perovskite materials has been designed and powered employing three different TENGs (GO paper/ Kapton, FTO/Kapton and hand/ FTO). In the first step for circuit designing to have better performance of the photodetector in coupling with TENG, the effect load resistance amount in the circuit on the impedance matching the TENG and the inner resistance of the photodetector, has been investigated through output current amplitude. The investigation, shows that to achieve the high amount of the photocurrent, the load resistance should be positioned in both critical zone of the out-put voltage of the TENG and the resistance range of high power density production of the TENG. In the second step, for investigation the effect of the dark resistance of the photodetector on out-put current of the self-powered photodetector, a device with very lower initial resistance (All-oxide Cu2O/ZnO photodetector) has been used with and without different load resistance in the circuit; in this regard, it is concluding that the initial resistance is too important to have proper design impedance matching circuit.
EPFL scientists have developed a digital model of the fruit fly, Drosophila melanogaster, that realistically simulates the movements of the animal. The twin is a big step towards reverse engineering the neuromechanical control of animal behavior, and developing bioinspired robots.
“We used two kinds of data to build NeuroMechFly,” says Professor Pavan Ramdya at EPFL’s School of Life Sciences. “First, we took a real fly and performed a CT scan to build a morphologically realistic biomechanical model. The second source of data were the real limb movements of the fly, obtained using pose estimation software that we’ve developed in the last couple of years that allow us to precisely track the movements of the animal.”
Ramdya’s group, working with the group of Professor Auke Ijspeert at EPFL’s Biorobotics Laboratory, has published a paper in Nature Methods showcasing the first ever accurate “digital twin” of the fly Drosophila melanogaster, dubbed “NeuroMechFly”.
Recent technological advances have enabled the creation of increasingly sophisticated sensors that can track movements and changes in real-world environments with remarkable levels of precision. Many engineers are now working to make these sensors thinner so that they can be embedded in a variety of devices, including robotic limbs and wearable devices.
Researchers at Hong Kong University of Science and Technology have recently developed a thin sensor for computer vision applications, which is based on a micro lens array (MLA). MLAs are 1D or 2D arrays comprising several small lenses, which are generally arranged in either squared or hexagonal patterns.
“In this study, we combined an old technology, a micro lens array, with vision-based tactile sensors,” Xia Chen, one of the researchers who carried out the study, told TechXplore. “This work builds on the work using the pinhole arrays to capture the image. We wanted to achieve a thin-format vision-based tactile sensor, as few studies so far focused on changing the imaging system of vison-based tactile sensor.”
The manipulation of electromagnetic waves and information has become an important part of our everyday lives. Intelligent metasurfaces have emerged as smart platforms for automating the control of wave-information-matter interactions without manual intervention. They evolved from engineered composite materials, including metamaterials and metasurfaces. As a society, we have seen significant progress in the development of metamaterials and metasurfaces of various forms and properties.
In a paper published in the journal eLight on May 6, 2022, Professor Tie Jun Cui of Southeast University and Professor Lianlin Li of Peking University led a research team to review intelligent metasurfaces. “Intelligent metasurfaces: Control, Communication and Computing” investigated the development of intelligent metasurfaces with an eye for the future.
This field has refreshed human insights into many fundamental laws. They have unlocked many novel devices and systems, like cloaking, tunneling, and holograms. Conventional structure-alone or passive metasurfaces has moved towards intelligent metasurfaces by integrating algorithms and nonlinear materials (or active devices).
A technological demonstration from China recently presented the power of super drones that track objects and people with high precision. The remote-powered vehicles, developed by scholars from Zhejiang University, were deployed into a thick bamboo forest to test their capabilities.
A video released by the researchers shows that the drones maneuvered effectively over the complex obstacles of the forest. The demonstration of the machines creeped out many audiences, as the precision and navigation of the drones exceeded far more than those of the technologies we see today.
Engineers from China developed what might be the most advanced drone swarm to date. Learn more about the autonomous machines and how they performed in tests.
The latest “machine scientist” algorithms can take in data on dark matter, dividing cells, turbulence, and other situations too complicated for humans to understand and provide an equation capturing the essence of what’s going on.
Despite rediscovering Kepler’s third law and other textbook classics, BACON remained something of a curiosity in an era of limited computing power. Researchers still had to analyze most data sets by hand, or eventually with Excel-like software that found the best fit for a simple data set when given a specific class of equation. The notion that an algorithm could find the correct model for describing any data set lay dormant until 2009, when Lipson and Michael Schmidt, roboticists then at Cornell University, developed an algorithm called Eureqa.
Their main goal had been to build a machine that could boil down expansive data sets with column after column of variables to an equation involving the few variables that actually matter. “The equation might end up having four variables, but you don’t know in advance which ones,” Lipson said. “You throw at it everything and the kitchen sink. Maybe the weather is important. Maybe the number of dentists per square mile is important.”
One persistent hurdle to wrangling numerous variables has been finding an efficient way to guess new equations over and over. Researchers say you also need the flexibility to try out (and recover from) potential dead ends. When the algorithm can jump from a line to a parabola, or add a sinusoidal ripple, its ability to hit as many data points as possible might get worse before it gets better. To overcome this and other challenges, in 1992 the computer scientist John Koza proposed “genetic algorithms,” which introduce random “mutations” into equations and test the mutant equations against the data. Over many trials, initially useless features either evolve potent functionality or wither away.