In 2008, the journalist Jonah Lehrer paid a visit to a lab in Lausanne, Switzerland, to profile Henry Markram, a world-renowned neuroscientist. Markram, a South African, had trained at a series of élite institutions in Israel, the United States, and Germany; in the nineties, he published foundational papers on neural connections and synaptic activity. Markram's work in the laboratory, which involved piercing neural membranes with what Lehrer described as an "invisibly sharp glass pipette," was known for its painstaking precision. Lehrer's visit, however, had been occasioned not by Markram's incremental contributions to the field--it's not easy to sell a colorful profile on the basis of such publications as "The neural code between neocortical pyramidal neurons depends on neurotransmitter release probability"--but by Markram's pivot, in the early two-thousands, to brain simulation. Neuroscience, Markram declaimed to Lehrer, had reached an impasse. Researchers had generated an enormous wealth of fine-grained data, but the marginal returns had begun to diminish.

The next breakthrough for artificial intelligence may help scientists design and print smarter cell structures that mimic human organs - thereby reducing trial-and-error and costs while contributing to new methods to fight disease and improve human health. These three-dimensional cell structures are called organoids - they can be designed to exhibit unique organ function and are used to conduct complex research on human tissue physiology, genetic diseases, organ-specific infectious diseases and cancer. Organoids can replicate development of native tissue in the architecture, function, cellular composition and transcriptional profile of almost any organ. Current methods of manufacturing organoids have yet to demonstrate consistent and robust extraction of mature organoids from renewable cells. This is where AI comes into the picture - designing and testing organoids utilizing computers rather than traditional lab methods.

Croquet Corporation, the startup that created the first open Metaverse operating system (OS), announces that its top executives will be speaking at the Ericsson Imagine Possible event in Santa Clara, CA, as well as AWE EU 2022 in Lisbon, Portugal. John Payne, CEO; David Smith, Co-Founder and CTO; and Vanessa Freudenberg, Co-founder and Chief Architect, will emphasize the importance of the Open Metaverse as a communication tool and collaborative experience - a place to augment human intellect and enable our ability to create, explore, understand and solve problems collaboratively. The Ericsson Imagine Possible event will bring together leading innovators and thought leaders from technology frontrunners and enterprises as they share their vision for the future - from Extended Reality (XR) and the Metaverse to the future of Enterprise connectivity. AWE (Augmented World Expo) is the world's leading AR VR conference and expo bringing together a mix of CEOs, CTOs, designers, developers, creative agencies, futurists, analysts, investors and top press in a unique opportunity to learn, inspire, partner and experience firsthand one of the most exciting industries of our times. Croquet recently won the AWE Auggie Award "Startup to Watch" award at AWE 2022 in Santa Clara earlier this year.

Artificial intelligence-driven software also helps researchers skip many repetitive … AI and Machine LearningMedTechArtificial Intelligence …

Many studies are increasingly focusing on frame interpolation, which synthesizes intermediate pictures between a pair of input frames. The refresh rate can be increased, or slow-motion videos can be created using temporal up-sampling. There's been a new application popping up recently. Due to the ease with which digital photography, individuals will often shoot several shots in rapid succession to find the best one, as they can now produce several images in a matter of seconds. Interpolating between these "near duplicates" reveals scene (and some camera) motion, frequently offering a more appealing sense of the event than any original photographs and presenting an interesting potential.

‘Data Science’ and ‘Machine Learning’ at IIT Mandi. Education / IANS / Oct 10, 2022, 18:14 IST. ‘Data Science’ and ‘Machine Learning’ at IIT Mandi.

In order to offer a customized script tool and inspire professional scriptwriters, we present VScript. It is a controllable pipeline that generates complete scripts, including dialogues and scene descriptions, as well as presents visually using video retrieval. With an interactive interface, our system allows users to select genres and input starting words that control the theme and development of the generated script. We adopt a hierarchical structure, which first generates the plot, then the script and its visual Figure 1: An example of the generated script (right) presentation. A novel approach is also introduced with its visual presentation (top left) from VScript. to plot-guided dialogue generation by Given the inputs, i.e., genre and starting words, a plot treating it as an inverse dialogue summarization. is generated, which guides the generation of a script The experiment results show that our approach consisting of a scene description and a dialogue. The outperforms the baselines on both automatic words highlighted in pink show the belongingness to and human evaluations, especially in the given genre (Sci-Fi).

Entity Linking (EL) is the gateway into Knowledge Bases. Recent advances in EL utilize dense retrieval approaches for Candidate Generation, which addresses some of the shortcomings of the Lookup based approach of matching NER mentions against pre-computed dictionaries. In this work, we show that in the domain of Tweets, such methods suffer as users often include informal spelling, limited context, and lack of specificity, among other issues. We investigate these challenges on a large and recent Tweets benchmark for EL, empirically evaluate lookup and dense retrieval approaches, and demonstrate a hybrid solution using long contextual representation from Wikipedia is necessary to achieve considerable gains over previous work, achieving 0.93 recall.

Real-time music accompaniment generation has a wide range of applications in the music industry, such as music education and live performances. However, automatic real-time music accompaniment generation is still understudied and often faces a trade-off between logical latency and exposure bias. In this paper, we propose SongDriver, a real-time music accompaniment generation system without logical latency nor exposure bias. Specifically, SongDriver divides one accompaniment generation task into two phases: 1) The arrangement phase, where a Transformer model first arranges chords for input melodies in real-time, and caches the chords for the next phase instead of playing them out. 2) The prediction phase, where a CRF model generates playable multi-track accompaniments for the coming melodies based on previously cached chords. With this two-phase strategy, SongDriver directly generates the accompaniment for the upcoming melody, achieving zero logical latency. Furthermore, when predicting chords for a timestep, SongDriver refers to the cached chords from the first phase rather than its previous predictions, which avoids the exposure bias problem. Since the input length is often constrained under real-time conditions, another potential problem is the loss of long-term sequential information. To make up for this disadvantage, we extract four musical features from a long-term music piece before the current time step as global information. In the experiment, we train SongDriver on some open-source datasets and an original \`aiSong Dataset built from Chinese-style modern pop music scores. The results show that SongDriver outperforms existing SOTA (state-of-the-art) models on both objective and subjective metrics, meanwhile significantly reducing the physical latency.

We characterize the class of image plane transformations which realize rigid camera motions and call these transformations `rigidity preserving'. In particular, 2D translations of pinhole images are not rigidity preserving. Hence, when using CNNs for 3D inference tasks, it can be beneficial to modify the inductive bias from equivariance towards translations to equivariance towards rigidity preserving transformations. We investigate how equivariance with respect to rigidity preserving transformations can be approximated in CNNs, and test our ideas on both 6D object pose estimation and visual localization. Experimentally, we improve on several competitive baselines.