Category Archives: Machine Learning
Podcast: NYU’s Kolm on transaction costs and machine learning – Risk.net
Most forms of post-trade transaction cost analysis only consider the price impact of completed orders. But ignoring partially filled orders which are all too common when trading produces a distorted measure of execution quality.
Depending on what methodologies are used, you might be off by 20% to 30%, relative to the true transaction cost, says Petter Kolm, professor of finance and director of the Mathematics in Finance masters program at NYUs Courant Institute of Mathematical Sciences, and our guest for this episode of Quantcast.
Kolms latest paperwith Nicholas Westray, a visiting researcher in financial machine learning at the Courant Institute, explores the so-called clean-up costs of trades, which they define as the opportunity cost attributed to the part of the order that is unfilled.
Most trading firms use ad hoc techniques to measure the cost of partial fills. The paper proposes a streamlined way to quantify clean-up costs that can be consistently applied to different trading strategies. The setup assumes the market behaves like a propagator model. This allows for the transaction costs of partially filled orders to be modelled as if they were fully executed, capturing the effects on drift of the security as well as the market impact of the trade.
In this podcast, Kolm also discusses his other research interests, including the applications of machine learning and its various branches in finance, one of whichis natural language processing (NLP). Kolm and his team have used NLP to gauge investor sentiment on individual stocks by harvesting signals from financial news. Their research has shown that there is indeed a connection between sentiment and the successive behaviour of the stock.
Kolm is also working on various applications of reinforcement learning, which is becomingincreasingly popular. He is, however, more cautious than other quants about it applicationin finance. While the technique is promising, he warns that prior applications such as the Alpha Go system developed by Googles Deepmind benefited from a large and stable database for training. In finance, quants only have a limited history of prices to work with. Reinforcement learning has had a bit of hype all the cool kids are doing it these days, but I think people are starting to understand and separate hype from reality, he says.
Kolm says his future projects will focus on the application of deep learning and reinforcement learning to optimal execution and the trading of American options, as well as the use of NLP to generate trading signals.
Index
00:00 Intro and transaction cost analysis
02:30 How costly are clean-up costs?
05:10 The problem of quantifying clean-up costs
10:32 Reading financial news with NLP
16:50 Reinforcement learning
22:40 Deep learning and limit order books
26:20 Teaching machine learning techniques
30:20 Future research projects
To hear the full interview, listen in the player above, or download. Future podcasts in our Quantcast series will be uploaded to Risk.net. You can also visit the main page here to access all tracks, or go to the iTunes store orGoogle Podcaststo listen and subscribe.
Now also available on Spotify.
Read the original post:
Podcast: NYU's Kolm on transaction costs and machine learning - Risk.net
Playing Catch-Up To The Early Adopters Of Analytics, AI And Machine Learning – Africa.com
By Sarthak Rohal, VP IT services at AlphaCodes
Early adopters of data analytics, Artificial Intelligence (AI) and Machine Learning (ML) tools have found themselves in a position of favour in todays rapidly accelerating digital world. Far from being a point of differentiation, these technologies have become imperative for survival. As businesses that have lagged behind struggle to play catch up, a trusted IT partner becomes a critical business asset, helping organisations adapt and thrive in a new world.
Better decisions, faster
The Covid-19 pandemic has affected all businesses around the world, and recovering from its effects will be a top priority for the remainder of 2021 and beyond. While some businesses struggle with the new reality, others have seen it as an opportunity to improve their data and analytical assets, operationalise, and update their processes.
The key to business success today is the ability to make better decisions faster. This all hinges on the ability to analyse data, put the analytics to work through AI, and then leverage technology to train algorithms that enhance the decision-making process using ML.
With the sheer volume of data available today, it is beyond human ability to gather, analyse and deliver insight in any meaningful timeframe. Crucially, adoption of AI/ML should not be seen as a replacement for human resources, but rather an augmentation of human ability.
The goal should be to use data and analytics to increase revenue, improve efficiency, and respond to customer/market trends, driving better decisions that create a competitive advantage.
Adapt or risk irrelevance
According toGartner, by the end of 2024, 75% of enterprises will operationalise AI, driving a fivefold increase in streaming data and analytics infrastructures.Grand View Researchstates that the global AI market size is expected to grow at a Compound Annual Growth Rate (CAGR) of 42.2% from 2020 to 2027. AMcKinseysurvey reports that, for financial year 2019, 66% of respondents agreed that adoption of AI/ML in their business has helped increase revenue, while 40% cited a decrease in costs with the adoption of AI/ML.
What this all means is that AI/ML is no longer a competitive advantage, but is necessary simply to keep pace with global business. However, it can be challenging to get right, as highlighted by aDeloitte reportthat states that somewhere in the region of 94% of enterprises face problems when it comes to implementing AI.
Getting the foundations right
Before implementing AI in data analytics, organisations need to look at their data and make sure that they have sufficient data points for the AI to process. Without enough data points, AI will inevitably be biased toward a certain outcome, which means it will not provide meaningful analytical insight.
Quality data is essential in reducing noise and bias in the data, which in turn is essential for more accurate outcomes. It also reduces the computational power required by analytics, and speeds the model training process for ML, if data is clean and relevant from the outset.
It is also important to implement AI in the right place. Not everything needs AI to solve a problem, and an indiscriminate approach will reduce both value and impact. Additionally, organisations need to manage the change to maximise adoption and reduce the amount of confusion that may occur.
Partnerships to success
The biggest issue is that using AI in data analytics is not just about applying AI models to the data. It also needs an understanding of the data being captured for the analytics purpose while understanding which models would yield the best results. These are not skills that many enterprises necessarily possess in-house, which is why partnering with a reputable technology provider is key.
Maximising value from AI requires enterprises to focus their efforts on the right business lines with the right AI models. An experienced partner can help organisations understand the nuances of data and assist with gaining meaningful insights to drive business capabilities to a competitive advantage. In addition, a technology partner can help organisations understand which areas of business could optimally benefit from the use of AI.
AI/ML-related solutions will always provide an edge to business decision-makers, as they can simulate thousands of models and iterations and understand the risk and returns from each iteration, which is practically impossible without these next-generation technologies. Playing catch-up is now vital, as organisations that have not yet jumped on the AI/ML bandwagon will find it increasingly difficult to remain competitive.
Go here to read the rest:
Playing Catch-Up To The Early Adopters Of Analytics, AI And Machine Learning - Africa.com
Avalo uses machine learning to accelerate the adaptation of crops to climate change – TechCrunch
Climate change is affecting farming all over the world, and solutions are seldom simple. But if you could plant crops that resisted the heat, cold or drought instead of moving a thousand miles away, wouldnt you? Avalo helps plants like these become a reality using AI-powered genome analysis that can reduce the time and money it takes to breed hardier plants for this hot century.
Founded by two friends who thought theyd take a shot at a startup before committing to a life of academia, Avalo has a very direct value proposition, but it takes a bit of science to understand it.
Big seed and agriculture companies put a lot of work into creating better versions of major crops. By making corn or rice ever so slightly more resistant to heat, insects, drought or flooding, they can make huge improvements to yields and profits for farmers, or alternatively make a plant viable to grow somewhere it couldnt before.
There are big decreases to yields in equatorial areas and its not that corn kernels are getting smaller, said co-founder and CEO Brendan Collins. Farmers move upland because salt water intrusion is disrupting fields, but they run into early spring frosts that kill their seedlings. Or they need rust resistant wheat to survive fungal outbreaks in humid, wet summers. We need to create new varieties if we want to adapt to this new environmental reality.
To make those improvements in a systematic way, researchers emphasize existing traits in the plant; this isnt about splicing in a new gene but bringing out qualities that are already there. This used to be done by the simple method of growing several plants, comparing them, and planting the seeds of the one that best exemplifies the trait like Mendel in Genetics 101.
Nowadays, however, we have sequenced the genome of these plants and can be a little more direct. By finding out which genes are active in the plants with a desired trait, better expression of those genes can be targeted for future generations. The problem is that doing this still takes a long time as in a decade.
The difficult part of the modern process stems (so to speak) from the issue that traits, like survival in the face of a drought, arent just single genes. They may be any number of genes interacting in a complex way. Just as theres no single gene for becoming an Olympic gymnast, there isnt one for becoming drought-resistant rice. So when the companies do what are called genome-wide association studies, they end up with hundreds of candidates for genes that contribute to the trait, and then must laboriously test various combinations of these in living plants, which even at industrial rates and scales takes years to do.
Numbered, genetically differentiated rice plants being raised for testing purposes. Image Credits: Avalo
The ability to just find genes and then do something with them is actually pretty limited as these traits become more complicated, said Mariano Alvarez, co-founder and CSO of Avalo. Trying to increase the efficiency of an enzyme is easy, you just go in with CRISPR and edit it but increasing yield in corn, there are thousands, maybe millions of genes contributing to that. If youre a big strategic [e.g., Monsanto] trying to make drought-tolerant rice, youre looking at 15 years, 200 million dollars its a long play.
This is where Avalo steps in. The company has built a model for simulating the effects of changes to a plants genome, which they claim can reduce that 15-year lead time to two or three and the cost by a similar ratio.
The idea was to create a much more realistic model for the genome thats more evolutionarily aware, said Collins. That is, a system that models the genome and genes on it that includes more context from biology and evolution. With a better model, you get far fewer false positives on genes associated with a trait, because it rules out far more as noise, unrelated genes, minor contributors and so on.
He gave the example of a cold-tolerant rice strain that one company was working on. A genomewide association study found 566 genes of interest, and to investigate each costs somewhere in the neighborhood of $40,000 due to the time, staff and materials required. That means investigating this one trait might run up a $20 million tab over several years, which naturally limits both the parties who can even attempt such an operation, and the crops that they will invest the time and money in. If you expect a return on investment, you cant spend that kind of cash improving a niche crop for an outlier market.
Were here to democratize that process, said Collins. In that same body of data relating to cold-tolerant rice, We found 32 genes of interest, and based on our simulations and retrospective studies, we know that all of those are truly causal. And we were able to grow 10 knockouts to validate them, three in a three-month period.
In each graph, dots represent confidence levels in genes that must be tested. The Avalo model clears up the data and selects only the most promising ones. Image Credits: Avalo
To unpack the jargon a little there, from the start Avalos system ruled out more than 90% of the genes that would have had to be individually investigated. They had high confidence that these 32 genes were not just related, but causal having a real effect on the trait. And this was borne out with brief knockout studies, where a particular gene is blocked and the effect of that studied. Avalo calls its method gene discovery via informationless perturbations, or GDIP.
Part of it is the inherent facility of machine learning algorithms when it comes to pulling signal out of noise, but Collins noted that they needed to come at the problem with a fresh approach, letting the model learn the structures and relationships on its own. And it was also important to them that the model be explainable that is, that its results dont just appear out of a black box but have some kind of justification.
This latter issue is a tough one, but they achieved it by systematically swapping out genes of interest in repeated simulations with what amount to dummy versions, which dont disrupt the trait but do help the model learn what each gene is contributing.
Avalo co-founders Mariano Alvarez (left) and Brendan Collins by a greenhouse. Image Credits: Avalo
Using our tech, we can come up with a minimal predictive breeding set for traits of interest. You can design the perfect genotype in silico [i.e., in simulation] and then do intensive breeding and watch for that genotype, said Collins. And the cost is low enough that it can be done by smaller outfits or with less popular crops, or for traits that are outside possibilities since climate change is so unpredictable, who can say whether heat- or cold-tolerant wheat would be better 20 years from now?
By reducing the capital cost of undertaking this exercise, we sort of unlock this space where its economically viable to work on a climate-tolerant trait, said Alvarez.
Avalo is partnering with several universities to accelerate the creation of other resilient and sustainable plants that might never have seen the light of day otherwise. These research groups have tons of data but not a lot of resources, making them excellent candidates to demonstrate the companys capabilities.
The university partnerships will also establish that the system works for fairly undomesticated plants that need some work before they can be used at scale. For instance it might be better to supersize a wild grain thats naturally resistant to drought instead of trying to add drought resistance to a naturally large grain species, but no one was willing to spend $20 million to find out.
On the commercial side, they plan to offer the data handling service first, one of many startups offering big cost and time savings to slower, more established companies in spaces like agriculture and pharmaceuticals. With luck Avalo will be able to help bring a few of these plants into agriculture and become a seed provider as well.
The company just emerged from the IndieBio accelerator a few weeks ago and has already secured $3 million in seed funding to continue their work at greater scale. The round was co-led by Better Ventures and Giant Ventures, with At One Ventures, Climate Capital, David Rowan and of course IndieBio parent SOSV participating.
Brendan convinced me that starting a startup would be way more fun and interesting than applying for faculty jobs, said Alvarez. And he was totally right.
Originally posted here:
Avalo uses machine learning to accelerate the adaptation of crops to climate change - TechCrunch
Taktile makes it easier to leverage machine learning in the financial industry – TechCrunch
Meet Taktile, a new startup that is working on a machine learning platform for financial services companies. This isnt the first company that wants to leverage machine learning for financial products. But Taktile wants to differentiate itself from competitors by making it way easier to get started and switch to AI-powered models.
A few years ago, when you could read machine learning and artificial intelligence in every single pitch deck, some startups chose to focus on the financial industry in particular. It makes sense as banks and insurance companies gather a ton of data and know a lot of information about their customers. They could use that data to train new models and roll out machine learning applications.
New fintech companies put together their own in-house data science team and started working on machine learning for their own products. Companies like Younited Credit and October use predictive risk tools to make better lending decisions. They have developed their own models and they can see that their models work well when they run them on past data.
But what about legacy players in the financial industry? A few startups have worked on products that can be integrated in existing banking infrastructure. You can use artificial intelligence to identify fraudulent transactions, predict creditworthiness, detect fraud in insurance claims, etc.
Some of them have been thriving, such as Shift Technology with a focus on insurance in particular. But a lot of startups build proof-of-concepts and stop there. Theres no meaningful, long-term business contract down the road.
Taktile wants to overcome that obstacle by building a machine learning product that is easy to adopt. It has raised a $4.7 million seed round led by Index Ventures with Y Combinator, firstminute Capital, Plug and Play Ventures and several business angels also participating.
The product works with both off-the-shelf models and customer-built models. Customers can customize those models depending on their needs. Models are deployed and maintained by Taktiles engine. It can run in a customers cloud environment or as a SaaS application.
After that, you can leverage Taktiles insights using API calls. It works pretty much like integrating any third-party service in your product. The company tried to provide as much transparency as possible with explanations for each automated decision and detailed logs. As for data sources, Taktile supports data warehouses, data lakes as well as ERP and CRM systems.
Its still early days for the startup, and its going to be interesting to see whether Taktiles vision pans out. But the company has already managed to convince some experienced backers. So lets keep an eye on them.
Visit link:
Taktile makes it easier to leverage machine learning in the financial industry - TechCrunch
Improve Machine Learning Performance by Dropping the Zeros – ELE Times
KAUST researchers have found a way to significantly increase the speed of training. Large machine learning models can be trained significantly faster by observing how frequently zero results are produced in distributed machine learning that uses large training datasets.
AI models develop their intelligence by being trained on datasets that have been labelled to tell the model how to differentiate between different inputs and then respond accordingly. The more labelled data that goes in, the better the model becomes at performing whatever task it has been assigned to do. For complex deep learning applications, such as self-driving vehicles, this requires enormous input datasets and very longtrainingtimes, even when using powerful and expensive highly parallel supercomputing platforms.
During training, small learning tasks are assigned to tens or hundreds of computing nodes, which then share their results over acommunications networkbefore running the next task. One of the biggest sources of computing overhead in such parallel computing tasks is actually this communication among computing nodes at each model step.
Communication is a major performance bottleneck in distributed deep learning, explains the KAUST team. Along with the fast-paced increase in model size, we also see an increase in the proportion of zero values that are produced during the learning process, which we call sparsity. Our idea was to exploit this sparsity to maximize effective bandwidth usage by sending only non-zero data blocks.
Building on an earlier KAUST development called SwitchML, which optimized internode communications by running efficient aggregation code on the network switches that process data transfer, Fei, Marco Canini and their colleagues went a step further by identifying zero results and developing a way to drop transmission without interrupting the synchronization of the parallel computing process.
Exactly how to exploit sparsity to accelerate distributed training is a challenging problem, says the team. All nodes need to process data blocks at the same location in a time slot, so we have to coordinate the nodes to ensure that only data blocks in the same location are aggregated. To overcome this, we created an aggregator process to coordinate the workers, instructing them on which block to send next.
The team demonstrated their OmniReduce scheme on a testbed consisting of an array of graphics processing units (GPU) and achieved an eight-fold speed-up for typicaldeep learningtasks.
Read more from the original source:
Improve Machine Learning Performance by Dropping the Zeros - ELE Times
New imaging, machine-learning methods speed effort to reduce crops’ need for water – University of Illinois News
CHAMPAIGN, Ill. Scientists have developed and deployed a series of new imaging and machine-learning tools to discover attributes that contribute to water-use efficiency in crop plants during photosynthesis and to reveal the genetic basis of variation in those traits.
The findings are described in a series of four research papers led by University of Illinois Urbana-Champaign graduate students Jiayang (Kevin) Xie and Parthiban Prakash, and postdoctoral researchers John Ferguson, Samuel Fernandes and Charles Pignon.
The goal is to breed or engineer crops that are better at conserving water without sacrificing yield, said Andrew Leakey, a professor of plant biology and of crop sciences at the University of Illinois Urbana-Champaign, who directed the research.
Drought stress limits agricultural production more than anything else, Leakey said. And scientists are working to find ways to minimize water loss from plant leaves without decreasing the amount of carbon dioxide the leaves take in.
Plants breathe in carbon dioxide through tiny pores in their leaves called stomata. That carbon dioxide drives photosynthesis and contributes to plant growth. But the stomata also allow moisture to escape in the form of water vapor.
A new approach to analyzing the epidermis layer of plant leaves revealed that the size and shape of the stomata (lighter green pores) in corn leaves strongly influence the crops water-use efficiency.
Micrograph by Jiayang (Kevin) Xie
Edit embedded media in the Files Tab and re-insert as needed.
The amount of water vapor and carbon dioxide exchanged between the leaf and atmosphere depends on the number of stomata, their size and how quickly they open or close in response to environmental signals, Leakey said. If rainfall is low or the air is too hot and dry, there can be insufficient water to meet demand, leading to reduced photosynthesis, productivity and survival.
To better understand this process in plants like corn, sorghum and grasses of the genus Setaria, the team analyzed how the stomata on their leaves influenced plants water-use efficiency.
We investigated the number, size and speed of closing movements of stomata in these closely related species, Leakey said. This is very challenging because the traditional methods for measuring these traits are very slow and laborious.
For example, determining stomatal density previously involved manually counting the pores under a microscope. The slowness of this method means scientists are unable to analyze large datasets, Leakey said.
There are a lot of features of the leaf epidermis that normally dont get measured because it takes too much time, he said. Or, if they get measured, its in really small experiments. And you cant discover the genetic basis for a trait with a really small experiment.
To speed the work, Xie took a machine-learning tool originally developed to help self-driving cars navigate complex environments and converted it into an application that could quickly identify, count and measure thousands of cells and cell features in each leaf sample.
Jiayang (Kevin) Xie converted a machine-learning tool originally designed to help self-driving cars navigate complex environments into an application that can quickly analyze features on the surface of plant leaves.
Photo by L. Brian Stauffer
Edit embedded media in the Files Tab and re-insert as needed.
To do this manually, it would take you several weeks of labor just to count the stomata on a seasons worth of leaf samples, Leakey said. And it would take you months more to manually measure the sizes of the stomata or the sizes of any of the other cells.
The team used sophisticated statistical approaches to identify regions of the genome and lists of genes that likely control variation in the patterning of stomata on the leaf surface. They also used thermal cameras in field and laboratory experiments to quickly assess the temperature of leaves as a proxy for how much water loss was cooling the leaves.
This revealed key links between changes in microscopic anatomy and the physiological or functional performance of the plants, Leakey said.
By comparing leaf characteristics with the plants water-use efficiency in field experiments, the researchers found that the size and shape of the stomata in corn appeared to be more important than had previously been recognized, Leakey said.
Along with the identification of genes that likely contribute to those features, the discovery will inform future efforts to breed or genetically engineer crop plants that use water more efficiently, the researchers said.
The new approach provides an unprecedented view of the structure and function of the outermost layer of plant leaves, Xie said.
There are so many things we dont know about the characteristics of the epidermis, and this machine-learning algorithm is giving us a much more comprehensive picture, he said. We can extract a lot more potential data on traits from the images weve taken. This is something people could not have done before.
Leakey is an affiliate of the Carl R. Woese Institute for Genomic Biology at the U. of I. He and his colleagues report their findings in a study published in The Journal of Experimental Botany and in three papers in the journal Plant Physiology (see below).
The National Science Foundation Plant Genome Research Program, the Advanced Research Projects Agency-Energy, the Department of Energy Biosystems Design Program, the Foundation for Food and Agriculture Research Graduate Student Fellows Program, The Agriculture and Food Research Initiative from the U.S. Department of Agriculture National Institute of Food and Agriculture, and the U. of I. Center for Digital Agriculture supported this research.
The Best Of Our Knowledge #1614: The Rise Of The Machines – WAMC
Today we think nothing of seeing laptops and iPads in the classroom. But there have been attempts at creating so-called teaching machines since the early 20th Century. And its the history of those early teaching machines that Audrey Watters explores in her new book called Teaching Machines The History of Personalized Learning."
Audrey Watters is an education technology writer and creator of the blog Hack Education.
So after a discussion about the history of learning machines, we thought it would be a good idea to take another look at machine learning. Its a very different thing. Machine learning and "Artificial Intelligence are two terms that were coined in the 1950s but are only now beginning to be put to solving practical problems. In the past few years, machine learning algorithms have been used to automate the interpretation and analysis of clinical chemistry data in a variety of situations in the lab. In the September 2020 issue of the journal Clinical Chemistry, there is a paper on a machine learning approach for the automated interpretation of amino acid profiles in human plasma. The same issue contains an accompanying editorial titled Machine Learning for the Biochemical Genetics Laboratory. One of the authors of the editorial is Dr. Stephen Master, Chief of the Division of Laboratory Medicine at the Childrens Hospital of Philadelphia and an Associate Professor of Pathology and Laboratory Medicine at the Perelman School of Medicine of the University of Pennsylvania. I asked Dr. Master, first of all, what exactly is machine learning, and why would it be significant for the clinical laboratory?
Okay, so weve done some deep dives into teaching machines and machine learning, lets go for the hat trick and take on virtual reality. Thats the topic of todays Academic Minute.
See the original post here:
The Best Of Our Knowledge #1614: The Rise Of The Machines - WAMC
Some of the emerging AI And machine Learning trends of 2021 – Floridanewstimes.com
From consumer electronics and smart personal assistants, advanced quantum computing systems to leading-edge medical diagnostic systems artificial Intelligence and machine learning technologies are increasingly finding their way into everything as they have been hot topics in 2020. According to market researcher IDC, up 12.3 percent from 2019 revenue generated by AI hardware, software and services is expected to reach $156.5 billion worldwide this year. But when it comes to trends in the development and use of AI and ML technologies can be easy to lose sight of the forest for the trees. You should look at hown AI and machine learning are being developed and the ways they are being used not just in the types of applications they are finding their way into as we approach the end of a turbulent 2020.
The growth of AI And Machine Learning in Hyperautomation
Hyperautomationis the idea that most anything within an organization that can be automated such as legacy business processes should be automated which is identified by market research firm Gartner. Also known as digital process automation and intelligent process automation,the pandemic has advanced adoption of the concept. The major drivers of hyper automation are AI and machine learning which are the key components. On static packaged software, hyper-automation initiatives cannot rely to be successful. To changing occurrences and answer to unexpected situations, the automated business processes must be able to adapt. To allow the system to automatically improve over time and respond to changing business processes and requirements along with data generated by the automated system the AI, machine learning models and deep learning technology arrive, using learning algorithms and models. You can check out embedded hardware design services of Integra Sources where you may find out more about this topic and also can apply to work for them.
Through AI EngineeringBringing Discipline to AI Development
According to Gartners research, the percentage of AI projects which successfully make it from prototype to full production is only about 53 percent. AI initiatives often fail to generate the hoped-for returns because businesses and organizations often struggle with system maintainability, scalability and governance when trying to deploy newly developed AI systems and machine learning models. According to Gartners list of Top Strategic Technology Trends for 2021 the performance, scalability, interpretability and reliability of AI models and deliver the full value of AI investments, will improve due to Businesses and organizations which are coming to understand the robust AI engineering strategy.
Link:
Some of the emerging AI And machine Learning trends of 2021 - Floridanewstimes.com
Bodo.ai Raises $14 million Series A to Revolutionize Simplicity, Performance and Scale for Data Analytics and Machine Learning – Business Wire
SAN FRANCISCO--(BUSINESS WIRE)--Bodo.ai, the extreme-performance parallel compute platform for data workloads, today announced it has raised $14 million in Series A funding led by Dell Technologies Capital, with participation from Uncorrelated Ventures, Fusion Fund and Candou Ventures.
Founded in 2019 to revolutionize complex data analytics and machine learning applications, Bodos goal is to make Python a first-class, high-performance and production-ready platform. The companys innovative compiler technology enables customers to solve challenging, large-scale data and machine learning problems at extreme performance and low cost with the simplicity and flexibility of native Python. Validated at 10,000+ cores and petabytes of data, Bodo delivers a previously unattainable supercomputing-like performance with linear parallel scalability. By eliminating the need to use new libraries or APIs or rewrite Python into Scala, C++, Java, or GPU code to achieve scalability, Bodo users may achieve a new level of performance and economic efficiency for large-scale ETL, Data Prep, Feature Engineering, and AI/ML Model training.
Big data is getting bigger, and in todays data-driven economy, enterprise customers need speed and scale for their data analytics needs, said Behzad Nasre, co-founder and CEO of Bodo.ai. Existing workarounds for large scale data processing like extra libraries and frameworks fail to address the underlying scale and performance issues. Bodo not only addresses this, but does so with an approach that requires no rewriting of the original application code.
Python is the second most popular programming language in existence largely due to its popularity among AI and ML developers and data scientists. However, most developers and data engineers who rely on Python for AI and ML algorithms are hampered by its sub-optimal performance when handling large-scale data. And those who use extensions and frameworks still find their performance falls orders of magnitude short of Bodos. For example, a large retailer recently achieved more than 100x real time performance improvement for their mission-critical program metric analysis workloads and saved over 90% on cloud infrastructure costs by using Bodo as opposed to a leading cloud data platform.
Customers know that parallel computing is the only way to keep up with computational demands for artificial intelligence and machine learning and extend Moores Law. But such high-performance computing has only been accessible to select experts at large tech companies and government laboratories, added Ehsan Totoni, co-founder and CTO of Bodo.ai. Our inferential compiler technology automates the parallelization formerly done by performance experts, democratizing compute power for all developers and enterprises. This will have a profound impact on large-scale AI, ML and analytics communities.
Bodo bridges the simplicity-vs-performance gap by delivering compute performance and runtime efficiency with no application rewriting. This will enable hundreds of thousands of Python developers and data scientists to perform near-real-time analytics and unlock new revenue opportunities for customers.
We see enterprises using more ML and data analytics to drive business insight and growth. There is a nearly constant need for more and better insights at near-real-time, said Daniel Docter, Managing Director, Dell Technologies Capital. But the exploding growth in data and analytics comes with huge hidden costs - massive infrastructure spend, code rewriting, complexity, and time. We see Bodo attacking these problems head-on, with an elegant approach that works for native Python for scale-out parallel processing. It will change the face of analytics.
For more information visit http://www.bodo.ai.
About Bodo.ai
Founded in 2019, Bodo.ai is an extreme-performance parallel compute platform for data analytics, scaling past 10,000 cores and petabytes of data with unprecedented efficiency and linear scaling. Leveraging unique automatic parallelization and the first inferential compiler, Bodo is helping F500 customers solve some of the worlds most massive data analysis problems. And doing so in a fraction of traditional time, complexity, and cost, all while leveraging the simplicity and flexibility of native Python. Developers can deploy Bodo on any infrastructure, from a laptop to a public cloud. Headquartered in San Francisco with offices in Pittsburgh, PA, the team of passionate technologists aims to radically accelerate the world of data analytics. http://bodo.ai #LetsBodo
About Dell Technologies Capital
Dell Technologies Capital is the global venture capital investment arm of Dell Technologies. The investment team backs passionate early stage founders who push the envelope on technology innovation for enterprises. Since inception in 2012, the team has sustained an investment pace of $150 million a year and has invested in more than 125 startups, 52 of which have been acquired and 7 have gone public. Portfolio companies also gain unique access to the go-to-market capabilities of Dell Technologies (Dell, Dell EMC, VMWare, Pivotal, Secureworks). Notable investments include Arista Networks, Cylance, Docusign, Graphcore, JFrog, MongoDB, Netskope, Nutanix, Nuvia, RedisLabs, RiskRecon, and Zscaler. Headquartered in Palo Alto, California, Dell Technologies Capital has offices in Boston, Austin, and Israel. For more information, visit http://www.delltechnologiescapital.com.
Go here to read the rest:
Bodo.ai Raises $14 million Series A to Revolutionize Simplicity, Performance and Scale for Data Analytics and Machine Learning - Business Wire
How AI and Machine Learning are changing the tech industry – refreshmiami.com
AI and Machine Learning have been gaining momentum over the past few years, but recently with the pandemic, it has accelerated in ways we couldnt imagine. Last year was an extremely difficult year for every imaginable sector of the economy. It has forced the acceleration of AI.
In this event, we will talk about how AI is changing the tech industry, and how the talent pool is not growing fast enough to meet the demands
Companies across all industries have been scrambling to secure top AI talent from a pool thats not growing fast enough. Even during the economic disruptions and layoffs caused by the COVID-19 pandemic, the demand for AI talent has been strong. Leaders are looking to reduce costs through automation and efficiency, and AI has a real role to play in that effort
Our panel will be comprised of amazing people in the industry
Koyuki Nakamori Head of Machine Learning at HeadSpace
Nehar Poddar Machine Learning Engineer at DEKA Research and Development
Excerpt from:
How AI and Machine Learning are changing the tech industry - refreshmiami.com