"TheBeatles: Get Back" is eight hours of carefully curated audio and footage from The Beatles in the studio and performing a rooftop concert in London in 1969. Jackson had to dig through 60 hours of vintage film footage and around 150 hours of audio recordings in order to put together his three-part documentary. Once he decided which footage and audio to include, then he had to take the next difficult step: cleaning up and restoring them both to give fans a look at TheBeatles like they had never seen them before.

In order to clean up the audio for "Get Back," Jackson employed algorithm technology to teach computers what different instruments and voices sounded like so they could isolate each track:

Once each track was isolated, sound mixers could then adjust volume levels individually to help with sound quality and clarity. The isolated tracks also make it much easier to remove noise from the audio tracks, like background sounds or the electronic hum of older recording equipment. This ability to fine-tune every aspect of the audio allowedJackson to make it sound like theFab Four are hanging out in your living room. When that technology is used for their musical performances, it's all the more impressive, as their rooftop concert feels as close to the real thing as you can possibly get.

Check out "TheBeatles: Get Back," streaming on Disney+.

Read this article:
The Beatles: Get Back Used High-Tech Machine Learning To Restore The Audio - /Film

If your company isn't treating data like an asset, youcould bemissing out ona majorgrowth opportunity.

That's according to SwamiSivasubramanian, vice president of Amazon Machine Learning.Sivasubramanianwas speaking duringa keynote conversation ondata and machine learning WednesdayatAWS re:Invent,a conference forbusiness owners and other technical decision-makers hosted byAmazon Web Services in Las Vegas.

Sivasubramanian says there are three thingscompaniescan do to make the most of their data. Here's his advice.

1.Modernize your data infrastructure.

Too many companies still treat their data like it's the 1990s when they should be implementing a modern data strategy, according to Sivasubramanian.This applies to both storing your data and"putting your data to work," he says. In many cases,hiring an outside company tomanage your databasefor you can save resources and help ensure your operations run smoothly. Sivasubramanian adds that acloud-basedsolution will help ensure that your company'sdata--even the most obscure, infrequently used bits--can be easily accessed by your teams that need it.

Applying modern solutions like machine learningto your database can alsohelp you detect problems faster. For example, an applicationslowdown that might otherwise go undetected for dayscan be identified and diagnosed quickly with machine learning. It can also provide suggestions for fixing problems with your data,which can be time consuming and costly if you're still doing somanually.

2.Unify your data.

It's important to have whatSivasubramanian refers to as a"single source of truth" about your business. Ensuring that your teams are all looking at the same data can help your company make the most of it. Of course,this doesn't mean every teamshould have access to every piece of data; different teams can and should have different permissions and levels of access. What's important is that this data is consistently reported and recorded.

"Opportunities to transform your business with data exist all along the value chain," says Sivasubramanian."But creating such a solution requires companies to have a full picture and a single view of their customers and their business."

3.Find innovative uses foryour data.

Applying insights to your data can help you improve existing operationsor build entirely new ones.Sivasubramanian pointsto severalAWS customers that have benefited from applying machine learning and analytics to their data.Tyson Foods has usedcameras armed with computer vision to identify ways to reducewaste bycutting down on packaging. AndPinterest has usednatural language processing to create more accurate search engines that allow employees to find the information they need faster.

"Machine learning is improving customer experiences, creating more efficiencies, and spurring completely new innovations," saysSivasubramanian. "And having the right data strategy is the key to these innovations."

Link:
AWS re:Invent: How to Use Machine Learning and Other Technology to Make the Most of Your Data - Inc.

Thyberg, J., Hedin, U., Sjlund, M., Palmberg, L. & Bottger, B. A. Regulation of differentiated properties and proliferation of arterial smooth muscle cells. Arteriosclerosis 10, 966990. https://doi.org/10.1161/01.ATV.10.6.966 (1990).

CAS Article PubMed Google Scholar

Owens, G. K., Kumar, M. S. & Wamhoff, B. R. Molecular regulation of vascular smooth muscle cell differentiation in development and disease. Physiol. Rev. 84, 767801. https://doi.org/10.1152/physrev.00041.2003 (2004).

CAS Article PubMed Google Scholar

Thyberg, J., Blomgren, K., Roy, J., Tran, P. K. & Hedin, U. Phenotypic modulation of smooth muscle cells after arterial injury is associated with changes in the distribution of laminin and fibronectin. J. Histochem. Cytochem. 45, 837846. https://doi.org/10.1177/002215549704500608 (1997).

CAS Article PubMed Google Scholar

Ross, R. & Glomset, J. A. Atherosclerosis and the arterial smooth muscle cell. Science 180, 13321339. https://doi.org/10.1126/science.180.4093.1332 (1973).

ADS CAS Article PubMed Google Scholar

Ross, R. Atherosclerosis: An inflammatory disease. N. Engl. J. Med. 340, 115126. https://doi.org/10.1056/nejm199901143400207 (1999).

CAS Article PubMed Google Scholar

Chappell, J. et al. Extensive proliferation of a subset of differentiated, yet plastic, medial vascular smooth muscle cells contributes to neointimal formation in mouse injury and atherosclerosis models. Circ. Res. 119, 13131323. https://doi.org/10.1161/circresaha.116.309799 (2016).

CAS Article PubMed PubMed Central Google Scholar

Michel, J.-B., Li, Z. & Lacolley, P. Smooth muscle cells and vascular diseases. Cardiovasc. Res. 95, 135137. https://doi.org/10.1093/cvr/cvs172 (2012).

CAS Article PubMed Google Scholar

Kothapalli, D. et al. Cardiovascular protection by ApoE and ApoE-HDL linked to suppression of ECM gene expression and arterial stiffening. Cell Rep. 2, 12591271. https://doi.org/10.1016/j.celrep.2012.09.018 (2012).

CAS Article PubMed PubMed Central Google Scholar

Liu, S.-L. et al. Cardiovascular protection in females linked to estrogen-dependent inhibition of arterial stiffening and macrophage MMP12. JCI Insight 4, e122742. https://doi.org/10.1172/jci.insight.122742 (2019).

Article PubMed Central Google Scholar

Klein, E. A. et al. Cell-cycle control by physiological matrix elasticity and in vivo tissue stiffening. Curr. Biol. 19, 15111518. https://doi.org/10.1016/j.cub.2009.07.069 (2009).

CAS Article PubMed PubMed Central Google Scholar

Bae, Y. H. et al. A FAK-Cas-Rac-lamellipodin signaling module transduces extracellular matrix stiffness into mechanosensitive cell cycling. Sci Signal 7, 57. https://doi.org/10.1126/scisignal.2004838 (2014).

CAS Article Google Scholar

Mui, K. L. et al. N-Cadherin induction by ECM stiffness and FAK overrides the spreading requirement for proliferation of vascular smooth muscle cells. Cell Rep. https://doi.org/10.1016/j.celrep.2015.02.023 (2015).

Article PubMed PubMed Central Google Scholar

Roden, D. M. et al. Opportunities and challenges in cardiovascular pharmacogenomics. Circ. Res. 122, 11761190. https://doi.org/10.1161/CIRCRESAHA.117.310965 (2018).

CAS Article PubMed PubMed Central Google Scholar

Johnson Julie, A. Ethnic differences in cardiovascular drug response. Circulation 118, 13831393. https://doi.org/10.1161/CIRCULATIONAHA.107.704023 (2008).

CAS Article PubMed PubMed Central Google Scholar

Roden, D. M. et al. Pharmacogenomics: Challenges and opportunities. Ann. Intern. Med. 145, 749757. https://doi.org/10.7326/0003-4819-145-10-200611210-00007 (2006).

Article PubMed PubMed Central Google Scholar

Donnelly, R. Characterizing variability in cardiovascular drug responses. Br. J. Clin. Pharmacol. 57, 535537. https://doi.org/10.1111/j.1365-2125.2004.02148.x (2004).

CAS Article PubMed Central Google Scholar

Carson, P., Ziesche, S., Johnson, G. & Cohn, J. N. Racial differences in response to therapy for heart failure: Analysis of the vasodilator-heart failure trials. Vasodilator-Heart Failure Trial Study Group. J. Cardiac Failure 5, 178187. https://doi.org/10.1016/s1071-9164(99)90001-5 (1999).

CAS Article Google Scholar

Wright, J. T. Jr. et al. Effect of blood pressure lowering and antihypertensive drug class on progression of hypertensive kidney disease: Results from the AASK trial. JAMA 288, 24212431. https://doi.org/10.1001/jama.288.19.2421 (2002).

CAS Article PubMed Google Scholar

Wright, J. T. Jr. et al. Outcomes in hypertensive black and nonblack patients treated with chlorthalidone, amlodipine, and lisinopril. JAMA 293, 15951608. https://doi.org/10.1001/jama.293.13.1595 (2005).

CAS Article PubMed Google Scholar

Agarwal, R. & Sinha, A. D. Cardiovascular protection with antihypertensive drugs in dialysis patients: Systematic review and meta-analysis. Hypertension 53, 860866. https://doi.org/10.1161/hypertensionaha.108.128116 (2009).

CAS Article PubMed Google Scholar

Genkel, V. V. & Shaposhnik, I. I. Conceptualization of heterogeneity of chronic diseases and atherosclerosis as a pathway to precision medicine: Endophenotype, endotype, and residual cardiovascular risk. Int. J. Chronic Dis. 2020, 5950813. https://doi.org/10.1155/2020/5950813 (2020).

Article PubMed PubMed Central Google Scholar

Yano, H. et al. Roles played by a subset of integrin signaling molecules in cadherin-based cell-cell adhesion. J. Cell Biol. 166, 283295. https://doi.org/10.1083/jcb.200312013 (2004).

CAS Article PubMed PubMed Central Google Scholar

Cheng, Z. et al. Focal adhesion kinase regulates smooth muscle cell recruitment to the developing vasculature. Arterioscler. Thromb. Vasc. Biol. 31, 21932202. https://doi.org/10.1161/atvbaha.111.232231 (2011).

CAS Article PubMed PubMed Central Google Scholar

Klein, E. A., Yung, Y., Castagnino, P., Kothapalli, D. & Assoian, R. K. in Methods in Enzymology Vol. Volume 426 (ed A. Cheresh David) 155175 (Academic Press, 2007).

Strauss, B. H. et al. Extracellular matrix remodeling after balloon angioplasty injury in a rabbit model of restenosis. Circ. Res. 75, 650658. https://doi.org/10.1161/01.res.75.4.650 (1994).

CAS Article PubMed Google Scholar

Braga, V. M., Machesky, L. M., Hall, A. & Hotchin, N. A. The small GTPases Rho and Rac are required for the establishment of cadherin-dependent cell-cell contacts. J. Cell Biol. 137, 14211431. https://doi.org/10.1083/jcb.137.6.1421 (1997).

CAS Article PubMed PubMed Central Google Scholar

Bae, Y. H., Liu, S. L., Byfield, F. J., Janmey, P. A. & Assoian, R. K. Measuring the stiffness of ex vivo mouse aortas using atomic force microscopy. JoVE https://doi.org/10.3791/54630 (2016).

Article PubMed PubMed Central Google Scholar

Bade, N. D., Kamien, R. D., Assoian, R. K. & Stebe, K. J. Curvature and Rho activation differentially control the alignment of cells and stress fibers. Sci. Adv. 3, e1700150. https://doi.org/10.1126/sciadv.1700150 (2017).

ADS CAS Article PubMed PubMed Central Google Scholar

Welsh, C. F. et al. Timing of cyclin D1 expression within G1 phase is controlled by Rho. Nat. Cell Biol. 3, 950957. https://doi.org/10.1038/ncb1101-950 (2001).

CAS Article PubMed Google Scholar

Nelson, C. M., Pirone, D. M., Tan, J. L. & Chen, C. S. Vascular endothelial-cadherin regulates cytoskeletal tension, cell spreading, and focal adhesions by stimulating RhoA. Mol. Biol. Cell 15, 29432953. https://doi.org/10.1091/mbc.e03-10-0745 (2004).

CAS Article PubMed PubMed Central Google Scholar

Shibata, R. et al. Role of rho-associated kinase in neointima formation after vascular injury. Circulation 103, 284289. https://doi.org/10.1161/01.CIR.103.2.284 (2001).

CAS Article PubMed Google Scholar

Tang, D. D. & Gunst, S. J. The small GTPase Cdc42 regulates actin polymerization and tension development during contractile stimulation of smooth muscle. J. Biol. Chem. 279, 5172251728. https://doi.org/10.1074/jbc.M408351200 (2004).

CAS Article PubMed Google Scholar

Lv, J. et al. Endothelial Cdc42 deficiency impairs endothelial regeneration and vascular repair after inflammatory vascular injury. Respir. Res. 19, 27. https://doi.org/10.1186/s12931-018-0729-8 (2018).

CAS Article PubMed PubMed Central Google Scholar

Li, J. et al. Essential role of Cdc42 in cardiomyocyte proliferation and cell-cell adhesion during heart development. Dev. Biol. 421, 271283. https://doi.org/10.1016/j.ydbio.2016.12.012 (2017).

CAS Article PubMed Google Scholar

Zuo, X. et al. Spheroids of endothelial cells and vascular smooth muscle cells promote cell migration in hyaluronic acid and fibrinogen composite hydrogels. Research 2020, 8970480. https://doi.org/10.34133/2020/8970480 (2020).

ADS CAS Article PubMed PubMed Central Google Scholar

Pfisterer, L., Feldner, A., Hecker, M. & Korff, T. Hypertension impairs myocardin function: a novel mechanism facilitating arterial remodelling. Cardiovasc. Res. 96, 120129. https://doi.org/10.1093/cvr/cvs247%JCardiovascularResearch (2012).

CAS Article PubMed Google Scholar

Friedrich, J., Seidel, C., Ebner, R. & Kunz-Schughart, L. A. Spheroid-based drug screen: considerations and practical approach. Nat. Protoc. 4, 309324. https://doi.org/10.1038/nprot.2008.226 (2009).

CAS Article PubMed Google Scholar

Foty, R. A simple hanging drop cell culture protocol for generation of 3D spheroids. JoVE https://doi.org/10.3791/2720 (2011).

Article PubMed PubMed Central Google Scholar

Sata, M. et al. A mouse model of vascular injury that induces rapid onset of medial cell apoptosis followed by reproducible neointimal hyperplasia. J. Mol. Cell. Cardiol. 32, 20972104. https://doi.org/10.1006/jmcc.2000.1238 (2000).

CAS Article PubMed Google Scholar

Marx, S. O., Totary-Jain, H. & Marks, A. R. Vascular smooth muscle cell proliferation in restenosis. Circ. Cardiovasc. Interv. 4, 104111. https://doi.org/10.1161/circinterventions.110.957332 (2011).

CAS Article PubMed Google Scholar

Danuser, G. Computer vision in cell biology. Cell 147, 973978. https://doi.org/10.1016/j.cell.2011.11.001 (2011).

CAS Article PubMed Google Scholar

McQuin, C. et al. Cell Profiler 3.0: Next-generation image processing for biology. PLoS Biol. 16, e2005970. https://doi.org/10.1371/journal.pbio.2005970 (2018).

CAS Article PubMed PubMed Central Google Scholar

Lee, K. et al. Functional hierarchy of redundant actin assembly factors revealed by fine-grained registration of intrinsic image fluctuations. Cell Syst. 1, 3750. https://doi.org/10.1016/j.cels.2015.07.001 (2015).

CAS Article PubMed PubMed Central Google Scholar

Piccinini, F. AnaSP: a software suite for automatic image analysis of multicellular spheroids. Comput. Methods Programs Biomed. 119, 4352. https://doi.org/10.1016/j.cmpb.2015.02.006 (2015).

Article PubMed Google Scholar

Psycharakis, S. E. et al. in Clinical and Preclinical Optical Diagnostics II. 11076_11024 (Optical Society of America).

Piccinini, F., Tesei, A., Zanoni, M. & Bevilacqua, A. ReViMS: Software tool for estimating the volumes of 3-D multicellular spheroids imaged using a light sheet fluorescence microscope. Biotechniques 63, 227229. https://doi.org/10.2144/000114609 (2017).

CAS Article PubMed Google Scholar

Falk, T. et al. U-Net: Deep learning for cell counting, detection, and morphometry. Nat. Methods 16, 6770. https://doi.org/10.1038/s41592-018-0261-2 (2019).

CAS Article PubMed Google Scholar

Jang, D. et al. A deep learning-based segmentation pipeline for profiling cellular morphodynamics using multiple types of live cell microscopy. Cell Reports Methods. 1, 100105. https://doi.org/10.1016/j.crmeth.2021.100105 (2021).

Ronneberger, O., Fischer, P. & Brox, T. U-Net: Convolutional networks for biomedical image segmentation. arXiv:1505.04597 (2015). https://ui.adsabs.harvard.edu/abs/2015arXiv150504597R.

Simonyan, K. & Zisserman, A. Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556 (2014). https://ui.adsabs.harvard.edu/abs/2014arXiv1409.1556S.

Kim, S. J. et al. Deep transfer learning-based hologram classification for molecular diagnostics. Sci. Rep. 8, 17003. https://doi.org/10.1038/s41598-018-35274-x (2018).

ADS CAS Article PubMed PubMed Central Google Scholar

Iglovikov, V. & Shvets, A. TernausNet: U-Net with VGG11 Encoder pre-trained on imagenet for image segmentation. arXiv:1801.05746 (2018).

Yosinski, J., Clune, J., Bengio, Y. & Lipson, H. How transferable are features in deep neural networks? arXiv:1411.1792 (2014). https://ui.adsabs.harvard.edu/abs/2014arXiv1411.1792Y.

Cowan, L. Y. P. a. L. Y. P. a. S. J. H. a. C. L. G. a. J. D. Discriminability-based transfer between neural networks. Adv. Neural Inf. Process. Syst. 5 (1993).

More here:
A machine learning pipeline revealing heterogeneous responses to drug perturbations on vascular smooth muscle cell spheroid morphology and formation |...

DeepMind CEO Demis Hassabis had a big 2021.

In December 2020, we published a list of 10 predictions about the world of artificial intelligence in the year 2021.

With 2021 now coming to a close, lets revisit these predictions to see how things actually played out. There is much to learn from these retrospectives about the state and trajectory of AI today.

Outcome: Wrong

As of the beginning of this year, no autonomous vehicle company had ever gone public. 2021 is the year that that all changed.

TuSimple, Embark and Aurora have all debuted on public markets this year. Argo is deep in preparations to go public. Plus.ai and Pony.ai both announced SPAC deals this year (though Pony.ai has since shelved its plans). Credible rumors are swirling about upcoming public market debuts for other autonomous players.

But Waymo and Cruise are not included on that list.

Given that Waymo and Cruise are the most well-capitalized of all AV companies, it makes sense that they would not necessarily be the first ones to need to tap public markets for more capital.

Still, while our timing proved premature, we expect both of these companies to eventually be publicly traded.

Outcome: Wrong

Deepfakes, which just a couple years ago were an oddity on the fringes of the Internet, have thrust themselves into mainstream public consciousness in 2021.

From an Anthony Bourdain documentary to viral Tom Cruise clips, from a widely condemned new pornography app to a bizarre story about a cheerleaders vindictive mom in small-town America, deepfakes are rapidly becoming a part of our societal milieu.

But no deepfake has yet fooled large numbers of viewers and caused meaningful real-world damage in the realm of U.S. politics. Lets hope it stays that way in 2022.

Outcome: Right(ish)

Research activity in federated learning has indeed surged this year.

The number of academic research papers published on federated learning grew from 254 in 2018, to 1,340 in 2019, to 3,940 in 2020, according to Google Scholar. In 2021 that number jumped to 9,110, with four weeks still left in the year.

In last years predictions we specified that this number would surpass 10,000 in 2021hence the ish. This one may come down to the wire....

Outcome: Wrong

No multi-billion-dollar acquisitions occurred in the world of AI chips in 2021.

Instead, the leading AI chip startups all raised rounds at multi-billion-dollar valuations, making clear that they aspire not to get acquired but to become large standalone public companies.

In our predictions last December, we identified three startups in particular as likely acquisition targets. Of these: SambaNova raised a $670 million Series D at a $5 billion valuation in April; Cerebras raised a $250 million Series F at a $4 billion valuation last month; and Graphcore raised $220 million at a valuation close to $3 billion amid rumors of an upcoming IPO.

Other top AI chip startups like Groq and Untether AI also raised big funding rounds in 2021.

Outcome: Wrong

In 2021, none of the leading AI drug discovery startups was acquired by a pharma incumbent. Instead, just like the AI chip startups in the previous section, these companies raised record amounts of funding to challenge the incumbents head-on.

Several AI drug discovery players completed IPOs in 2021, making them among the earliest AI-first companies in the world to trade on public markets.

Recursion went public in April; Exscientia followed it in October. Insilico is slated to IPO soon. Insitro, XtalPi and a handful of other AI drug discovery players raised massive private rounds this year. For most of these competitors, the window for an acquisition has likely passed.

Outcome: Right

Finally, a prediction that we nailed!

For years, U.S. policymakers have been relatively inattentive to the strategic importance of artificial intelligence while more forward-thinking governments like China and Canada have rolled out detailed national strategies to position themselves as global AI leaders.

This changed in a big way in 2021, with an explosion of U.S. public policy activity related to AI. At the beginning of the year, Congress passed legislation to promote and coordinate AI research. Numerous additional AI-related bills have been introduced in both chambers of Congress this year. A dedicated White House group has been established to oversee the nations overall approach to AI. The U.S. military has gone into overdrive in its AI investments. In October, the Biden administration called for an AI Bill Of Rights for the American people. The list goes on.

It would be going too far to say that the U.S. government has established a cohesive national AI strategy. But in 2021, artificial intelligence rocketed to the forefront of Washingtons policy agenda.

Outcome: Right

In January 2021, less than a month after we published our predictions, Google announced that it had trained a model with 1.6 trillion parameters, making it the largest AI model ever built.

Now the question ishow big will these models get in 2022?

Outcome: Right(ish)

The crowded MLOps landscape has begun to consolidate in 2021. In several instances this year, large AI platforms have acquired smaller startups building tools and infrastructure for machine learning.

Probably the most noteworthy example came in July with DataRobots acquisition of Algorithmia, which had raised close to $40 million in venture capital funding.

Other examples include HPEs acquisition of Determined AI and DataRobots acquisition of decision.ai.

But there was less M&A activity in MLOps this year than we expected. In last years predictions, we listed 14 MLOps startups that we saw as potential acquisition targets. Of these, only oneAlgorithmiaended up being acquired. (Several others on that listWeights & Biases, Snorkel AI, OctoMLinstead raised rounds at monster valuations.)

Outcome: Right

Regulatory momentum for antitrust action against Big Tech has been building for years given the outsize influence that companies like Alphabet, Amazon and Facebook exert over the economy. But over the past year, antitrust regulators have increasingly refined their messaging by focusing on the structural advantages that these giants enjoy in AI. The jumping-off point, almost always, is the companies unrivaled data assets and aggressive data accumulation practices.

From recent Senate antitrust hearings to presidential Executive Orders, this theme of unfair data advantages translating into unfair AI advantages is becoming an increasingly important dimension of the Big Tech antitrust movement.

Last month, for instance, Lina Khans Federal Trade Commission appointed prominent AI critic Meredith Whittaker to a special role as the FTCs senior adviser on AI. As one industry observer put it: Whittaker's hiring is just the latest evidence of the FTCs attention on algorithms and algorithmic issues.

Outcome: Right

Of the predictions on last years list, this one was the most open-ended and least verifiable. Even so, plenty of developments in 2021 point to the continued emergence of biology as the most important and high-impact of all AI application areas.

AI is transforming drug discovery, with profound implications for the pharmaceutical industry and the future of human health. AI-discovered therapeutics are now in clinic; AI drug discovery startups are now trading on public markets.

DeepMinds landmark AlphaFold work, which was published in July, is a testament to the almost magical potential for machine learning to uncover fundamental truths about how life works. We have previously argued in this column that AlphaFold is the most important achievement in the history of AI. As Alphabets big announcement about Isomorphic Labs last month underscores, AlphaFold is just the beginning.

Perhaps more so than any other area of AI, world-class talent and investment dollars are flooding into computational biology. Take, for example, Eric Schmidts $150 million donation earlier this year to establish a new center at Harvard and MIT that will catalyze a new scientific discipline at the intersection of biology and machine learning.

In the years ahead, the application of computational methods and machine learning to biology is poised to transform societyand perhaps life as we know it.

Note: The author is a Partner at Radical Ventures, which is an investor in Untether AI.

Excerpt from:
2021 AI Predictions: What We Got Right And Wrong - Forbes

Webcast Combining Serverless and BERT for accuracy and cost-effectiveness with the MCubed web lecture series

An old truism of Machine Learning assumes that the more complex (and therefore the larger) a model is, the more accurate the outcome of its predictions. And indeed, if youre looking into machine learning disciplines like natural language processing (NLP), its the massive models generated using BERT or GPT that currently get practitioners swooning when it comes to precision.

Enthusiasm fades when it comes to productionising models, however, as their sheer size turns deployments into quite a struggle. Not to mention the cost of setting up and maintaining the infrastructure needed to make the step from research to production happen.

Reading this, avid followers of IT trends might now remember the emergence of Serverless Computing a couple of years ago. The approach pretty much promised large computing capabilities that could automatically scale up and down to satisfy changing demands and keep costs low. It also brought about an option to free teams from the burden of looking after their infrastructure, as it mostly came in the form of managed offerings.

Well, serverless hasnt gone anywhere since then, and seems like an almost ideal solution on first looks. Digging deeper however, limitations on things like memory occupation and deployment package size stand in the way of making it a straightforward option. Interest in combining serverless and machine learning is growing, though. And with it the number of people working on ways to make BERT models and Co fit provider specifications to facilitate serverless deployments.

To learn more about these developments, well welcome Marek uppa to episode 4 of our MCubed web lecture series for machine learning practitioners on December 2. uppa is head of data at Q&A and polling app Slido, where he and some colleagues used the last year to investigate ways to modify models for sentiment analysis and classification so that they can be used in serverless environments without dreaded performance degradations.

In his talk, uppa will speak a bit about his teams use case, the things that made them consider serverless, troubles they encountered during their studies, and the approaches they found to be the most promising to reach latency levels appropriate for production environments for their deployments.

As usual, the webcast on December 2 will start at 11:00 UTC with a roundup of software development-related machine learning news, which will give you a couple of minutes to settle in before we dive into the topic of model deployment in serverless environments. Wed love to see you there well even send you a quick reminder on the day, just register here.

And if machine learning at large still seems exciting but a bit out of reach for you, were sure our introductory online workshop with Prof Mark Whitehorn on December 9 can help you get started. Head over to the MCubed website for more information and tickets.

Here is the original post:
Skint but looking to get complex machine learning models into production? Serverless might be the answer DEVCLASS - DevClass

SEATTLE--(BUSINESS WIRE)--Today the Intelligent Applications Top 40 (#IA40) was unveiled by Madrona Venture Group. The list of the top private companies building applications that truly incorporate intelligence into how they process data and predict outcomes was voted on by the top 40 venture capital firms investing in this industry, Goldman Sachs, Amazon and Microsoft. http://www.ia40.com

The era of SaaS applications is drawing to a close as their intelligent application counterparts rise to prominence. We have been investing in intelligent apps since 2011 and we believe that in 2022, we will see a significant shift in how companies build applications, deliver insights and change our lives and these 40 companies will be driving that shift, commented Matt McIlwain, Managing Director, Madrona Venture Group. We are honored to partner with Goldman Sachs and all the top tier venture firms who contributed to this ranking of the companies that are building the future of software.

To create this list, Madrona Venture Group and Goldman Sachs collaborated on a new initiative to research, identify and select the top intelligent application companies in the private sector. The IA40 initiative enlisted judges across 40 of the most active venture capital firms to participate in this research-driven ranking. Judges nominated over 250 companies that they believe will transform and define the next generation of software with application intelligence, and then voted for the top 40 most promising.

The top 40 companies were selected across categories

Early

Mid

Growth

Enablers

Runway ML

Abnormal Security

Gong

dbt labs

Tesorio

Instabase

Snyk

Databricks

Spot AI

Loom

Samsara

Hugging Face

Auditoria

Amperity

Cribl

Cockroach Labs

TruEra

Primer.ai

Starburst Dat

DataRobot

Clockwise

Hyperscience

Celonis

Fivetran

WellSaid Labs

Axonius

Anduril

OctoML

LinearB

Cresta.ai

Chainalysis

Grafana Labs

Machinify

Seekout

Workato

Monte Carlo

Replicant

Moveworks

Snorkle AI

Synthesia

Top Line Insights

Insights

We thank all the judges of the IA 40.

a16z, Accel, Acrew Capital, Addition, Allen Institute of AI, Altimeter, Amazon, Amplify, Bain Capital Ventures, Battery Ventures, Bessemer Venture Partners, Coatue , CRV, Decibel Partners, defy.vc, Emergence Capital, General Catalyst, GGV Capital, Goldman Sachs, Greylock, ICONIQ Capital, Insight Partners, Khosla Ventures, Kleiner Perkins, Lux Capital, M12, Madrona Venture Group, March Capital, Mayfield Fund, Meritech Capital Partners, NEA, Norwest Venture Partners, OpenView, Positive Sum, Redpoint, Revolution Ventures, Scale Venture Partners, Sequoia Capital, STEADFAST Capital, Two Sigma Ventures.

Intelligent Application Definition

Intelligent applications leverage machine learning models embedded in applications that use both historical and real- time data to build a continuous learning system. These learning systems solve a business problem in a contextually relevant way - better than before, and typically deliver rich information and insights that are either applied automatically or leveraged by end users to make superior decisions.

About Madrona Venture Group

Madrona (www.madrona.com) is a venture capital firm based in Seattle, WA. With more than 25 years of investing in early stage technology companies, the firm has worked with founders from Day One to help build their company for the long run. Madrona invests predominantly in seed and Series A rounds across the information technology spectrum and has also raised Acceleration Stage funds for initial investments in Series B, C and beyond. Madrona was an early investor in companies such as Amazon, Smartsheet, Rover, Redfin, and Snowflake.

Read the original post:
Announcing the Intelligent Applications Top 40; a New Industry Ranking of the Top Private Companies Building Applications with AI and Machine Learning...

Mindtree has earned the AI and Machine Learning on Microsoft Azure advanced specialization, a validation of a services partners deep knowledge, extensive experience and proven success in enabling customer adoption of AI and implementing Azure solutions for machine learning life cycle and AI-powered apps.

Only partners that meet stringent criteria around customer success and staff skilling, as well as pass a third party audit of their AI and machine learning technical practices, can earn the AI and Machine Learning on Microsoft Azure advanced specialization.

Radhakrishnan Rajagopalan, Global Head, Customer Success, Data and Intelligence, Mindtree. Organizations are looking for ways to maximize business impact and revenue through augmentation and automation. As a result, AI and Machine Learning are playing an increasingly vital role in helping them unlock the full power of data for improved agility, richer experiences, smarter decision-making and reduced time-to-market. This advanced specialization validates our ability to enable organizations to optimize their digital strategies and investments, strengthening our reputation as a preferred digital transformation partner.

Rodney Clark, Corporate Vice President, Global Partner Solutions, Channel Sales and Channel Chief at Microsoft, added, AI and Machine Learning on Microsoft Azure advanced specialization highlights the partners who can be viewed as most capable when it comes to implementing Azure solutions for machine learning lifecycle and AI-powered apps. Mindtree clearly demonstrated that they have both the skills and the experience to enabling customer adoption of AI and Machine Learning in Microsoft Azure advanced specialization.

As the speed of business accelerates, organizations of every type and size are looking for ways to streamline processes and deliver simpler, faster, and smarter resources to help them keep up. Partners with the AI and Machine Learning on Microsoft Azure advanced specialization can give organizations the tools and knowledge to develop AI solutions on their terms, build AI into their mission-critical applications, and put responsible AI into action.

At around 12.46 pm, Mindtree is trading at Rs4431 per piece up by Rs55.55 or 1.27% on Sensex.

See original here:
Mindtree stock surges after earning the Al and Machine Learning on Microsoft Azure Advanced Specialization - Indiainfoline

key enablers | 19 November 2021 | Reporter

By: Reporter

DEWC Systems and the Australian Institute for Machine Learning (AIML) have agreed to partner on research to better detect radio signals in complex environments.

DEWC Systems and the Australian Institute for Machine Learning (AIML) have agreed to partner on research to better detect radio signals in complex environments.

DEWC Systems and the University of Adelaides Australian Institute for Machine Learning (AIML) have announced the commencement of a partnership to better understand how to apply artificial intelligence and machine learning to detect radio frequencies in difficult environments using MOESS and Wombat S3 technology.

As of yet, both organisations have already undertaken significant research on Phase 1 of the Miniaturised Orbital Electronic Sensor System (MOESS) project with the collaboration hoping to enhance the research yet further.

The original goal of the MOESS was to develop a platform to perform an array of applications and develop an automatic signal classification process. The Wombat 3 is a ground-based version of the MOESS.

Chief technology officer of DEWC Systems Dr Paul Gardner-Stephen will lead the project, which hopes to develop a framework for AI-enabled spectrum monitoring and automatic signal classification.

Radio spectrum is very congested, with a wide range of signals and interference sources, which can make it very difficult to identify and correctly classify the signals present. This is why we are turning to AI and ML, to bring the necessary algorithmic power necessary to solve this problem, Gardner-Stephen said.

"This will enable the creation of applications that work on DEWCs MOESS and Wombat S3 (Wombat Smart Sensor Suite) platforms to identify unexpected signals from among the forest of wireless communications, to help defence identify and respond to threats as they emerge.

According to Gardner-Stephen, both the MOESS and Wombat 3 platforms are highly capable software defined radio (SDR) platforms with on-board artificial intelligence and machine learning processors.

Since the project is oriented around creating an example framework, using two of DEWC Systems software defined radio (SDR) products, both DEWC Systems and AIML can create the kinds of improved situation awareness applications that use those features to generate the types of capabilities that will support defence in their mission, he explained.

In addition to directly working towards the creation of an important capability, it will also act to catalyse awareness of some of the kinds of applications that are possible with these platforms.

Subscribe to the Defence Connect daily newsletter. Be the first to hear the latest developments in the defence industry.

Chief executive of DEWC Systems Ian Spencer noted that the company innovates with academic institutions to develop leading technology.

Whilst we provide direction and guidance of the project, AIML will be bringing their deep understanding and cutting-edge technology of AI and machine learning. This is what DEWC Systems does. We collaborate with universities and other industry sectors to develop novel and effective solutions to support the ADO, Spencer said.

It is hoped that the technology developed throughout the partnership will support machine learning and artificial intelligence needs of Defence.

[Related:Veteran-owned SMEs DEWC Systems and J3Seven aim to solve mission critical challenges]

DEWC, AIML partner on AI and machine learning to enhance RF signal detection

Read more:
DEWC, AIML partner on AI and machine learning to enhance RF signal detection - Defence Connect

Event You want to know more about the ins and outs of machine learning, but cant figure out where to start? Our AI practitioners' conference MCubed and The Register regular Mark Whitehorn have got you covered.

Join us on December 9 for an interactive online workshop to learn all about ML types and algorithms, and find out about strengths and weaknesses of different approaches by using them yourself.

This limited one-day online workshop is geared towards anyone who wants to gain an understanding of machine learning no matter your background. Mark will start with the basics, asking and answering what is machine learning, before diving deeper into the different types of systems you keep hearing about.

Once youre familiar with supervised, unsupervised, and reinforcement learning, things will get hands-on with practical exercises using common algorithms such as clustering and, of course, neural networks.

In the process, youll also investigate the pros and cons of different approaches, which should help you in assessing what could work for a specific task and what isnt an option, and learn how the things youve just tried relate to what Big Biz are using. However, its not all code and algorithms in the world of ML, which is why Mark will also give you a taster of what else there is to think about when realizing machine learning projects, such as data sourcing, model training, and evaluation.

Since Python has turned into the language of choice for many ML practitioners, exercises and experiments will be performed in Python mostly, so installing it along with an IDE will help you make the most of the workshop if you havent already.

This doesnt mean the course is for Pythonistas only, however. If youre not familiar with the language, exercises will be transformed into demonstrations providing you insight into the inner workings of the associated code, before we start altering some of the parameters together. Like that, you get to find out how each parameter influences the learning that is performed, leaving you in top shape to continue in whatever language (or no-code ML system) you feel comfortable with.

Your trainer, Professor Mark Whitehorn, works as a consultant for national and international organisations, such as the Bank of England, Standard Life, and Sainsburys, designing analytical systems and data science solutions. He is also the Emeritus Professor of Analytics at the University of Dundee where he teaches a master's course in data science and conducts research into the development of analytical systems and proteomics. You can get a taster of his brilliant teaching skills here.

If this sounds interesting to you, head over to the MCubed website to secure your spot now. Tickets are very limited to make sure we can answer all your questions and everyone is getting proper support throughout the day so dont wait for too long.

Continued here:
MCubed does web workshops: Join Mark Whitehorns one-day introduction to machine learning next month - The Register

Commentary: The house always wins in gambling, and the house is getting even tougher through machine learning.

Image: iStock/Igor Kutyaev

"On the Internet, nobody knows you are a dog," is easily one of the top 10 New Yorker cartoons of all time. Why? Because it captured the upsides and downsides of online anonymity. All good, right? Well, maybe. What if you are online, and you like to gamble? Who's on the other side? You have no idea, and that might be more of a problem than you might suspect.

SEE: Artificial Intelligence Ethics Policy (TechRepublic Premium)

For one thing, more and more you may be betting against machine learning algorithms, and if the "house always wins" in the offline world, guess what? It's even worse in an ML/artificial intelligence-driven online gambling world. Still, understanding the odds helps you understand the potential risks involved as the gambling industry consolidates. So, let's take a look at how one person used ML to fight back.

Go to any casino in person and the best odds you can get range from the housetaking from 1.5% to 5% off the top (craps, baccarat, slot machines and Big Six can take more than 20%). You are essentially renting access to their game. The money you bet allows you to earn back about 95 to 98 cents on the dollar (the card game blackjack, by the way, is your best bet). But any way you choose, over time you almost certainly go broke. Why? Because ... math.

SEE: Research: Increased use of low-code/no-code platforms poses no threat to developers (TechRepublic Premium)

The casino industry willargue that AI/ML helps gamblers by identifying cheats faster. That might be true, so far as it goes, but there is another side to this argument.

I came across an intriguing example of a regular person using ML to see if they could do better at the racetrack betting on the ponies (a $15 billion annual industry in the U.S.). In this example, the regular person is Craig Smith, a noted former New York Times foreign correspondent who left journalism to explore AI/ML.

To test the efficacy of ML and horse racing, he tried Akkio, a no-code ML service I've written about a few times before. His goal? To show how their approach canfoster AI adoption and how it is alreadyimproving productivity in mundane but important matters. Akkio is not designed for gambling but rather for business analysts who want insights quickly into their data without hiring developers and data scientists. Turns out it's also helpful for Smith's purposes.

So much so, in fact, that Smithdoubled his money using an ML recommendation model Akkiocreated in minutes. It's a fascinating read. It also sheds light on the dark side of ML and gambling.

In his article, Smith interviewed Chris Rossi. He's the horse betting expert who helped build a thoroughbred data system that was eventually bought by the horse racing information conglomerate DRF (Daily Racing Form). He now consults for people in the horse-racing world, including what he described as teams of quantitative analysts who use machine learning to game the races betting billions annually and making big buckssome of it from volume rebates on losing bets by the tracks who encourage the practice.

"Horse racing gambling is basically the suckers against the quants," Rossi said. "And the quants are kicking the ---- out of the suckers."

Not many years ago, sports betting sat in a legally dubious place in the U.S. Then in 2018 the U.S. Supreme Court cleared the way for states tolegalize the practice, striking down a 1992 federal law that largely restricted gambling and sports books to Nevada. That decision arrived just in the nick of time. During the pandemic, as casinos shuttered their doors and consumers looked for activities to eat up their free time, online gambling and sports betting took off. Shares of DraftKings, which went public via a SPAC merger, for instance, have risen 350% since the start of the coronavirus' spread, valuing the company at about $22 billion.

SEE:Metaverse cheat sheet: Everything you need to know (free PDF)(TechRepublic)

DraftKings has also been looking to diversify away from business that concentrates around the sports season. The online betting customer is apparently more valuable than a sports betting customer.

More recently,MGM Resorts International, a major Las Vegas player, sought to acquire Entain for about $11.1 billion in January, though the latter rebuffed the bid for being too low. Caesars Entertainment in September announced plans to acquire U.K.-based online betting business, William Hill, for about $4 billion. And to drive the point home on just how hot the space has gotten, media brand Sports Illustrated has gotten into the online sports betting space.

All of this money sits awkwardly next to rising use of ML. Yes, ML can help clean up online gambling by kicking off cheaters. But it can also be the other side of the bet you are making. As one commentatornoted, "AI can analyze player behavior and create highly customized game suggestions." Such customized gaming may make it more engaging for gamblers to keep betting, but don't think for a minute that it will help them to win. Online or offline, the house always wins. If anything, the new ML-driven gambling future just means gamblers may have incentive to gamble longer and lose more.

Could you, like Smith, put ML to work on your behalf? Sure. But at some point, the house wins, and the house will improve its use of ML faster than any average bettor can.

Disclosure: I work for MongoDB, but the views expressed herein are mine.

Learn the latest news and best practices about data science, big data analytics, and artificial intelligence. Delivered Mondays

Link:
How machine learning is skewing the odds in online gambling - TechRepublic