Mediaboss Marketing

Hewlett Packard Enterprise Chief Technology Officer and Head of Software Kumar Sreekanti, the driving force behind the new HPE Ezmeral container platform and artificial intelligence machine-learning Ops software, said Ezmeral is the undisputed leader in the race to bring AI and ML to the enterprise.

HPE has built AI applications, said Sreekanti, a software innovator who started AI superstar BlueData, which was acquired by HPE in 2018. When it comes to the [AI] PaaS [Platform-as-a-Service] layer, we are the undisputed leader. Not only are we the undisputed leader but it is open source. So if somebody wants to run DataRobot they can. We have one customer who runs DataRobot on top of the Ezmeral platform. So we are the clear leader [with AI] in front compared to anybody. Tell me where a [Red Hat] OpenShift or a [VMware] Tanzu is running an AI ML application? We are in front.

As for AWS SageMaker AI offering, Sreekanti pointed out that it comes up short because it effectively prevents customers from moving seamlessly to another cloud. AWS sits in a walled garden with SageMaker, he said. The problem with AWS with SageMaker is that when you end up with SageMaker you are stuck with SageMaker. You cant lift it and go where you want. Whereas if you run on top of the BlueData Ezmeral software you can take that and run it somewhere else.

Sreekantiwho sometimes is referred to as the professor inside HPE, this week unveiled HPEs Ezmeral software platform, which essentially provides the IaaS {Infrastructure-as-a-Service] and PaaS platforms for partners and customers to build a new generation of intelligent applications that take advantage of the platforms robust AI and machine-learning capabilities.

Among HPEs big advantages with Ezmeral is the BlueData AI machine-learning software that Sreekanti delivered to customers at BlueData. That software is designed to run both stateful and stateless workloadsessentially allowing HPE to provide the same cloud experience in a cloud-native or on-premises workload in the data center or even in real time at the edge.

It is the only one that runs AI ML Spark [real-time data] workloads while running CICD [Continous Integration, Continuous Delivery], said Srkeenati.

The machine-learning operations for Ezmeral run with the Kubernetes open-source platform on the same hardware as the HPE MapR file system. What that does is give the customer the ability to get the best out of the whole organized stack from the top to the bottom, said Sreekanti.

Original post:
HPE Ezmeral Is 'Undisputed Leader' In AI Battle With AWS, VMware, Red Hat: Kumar Sreekanti - CRN: Technology news for channel partners and solution...

By Bio-IT World Staff

June 17, 2020 |Big investments this past month in machine-learning driven drug discovery frominsitro, andDNAnexussand cloud-based informatics platform. Vaccine funding forGreenLightBiosciences,boron drug delivery,and moreof the latest funding updates from across life sciences, clinical trials, the diagnostics industries.

$143M:Machine-Learning-Enabled Drug Discovery

insitro, a San Francisco machine-learning driven drug discovery and development company, has raised $143 million in an oversubscribed Series B financing. The financing was led by Andreessen Horowitz, with participation from new investors Canada Pension Plan Investment Board (CPP Investments) and funds and accounts advised by T. Rowe Price Associates, as well as funds managed by BlackRock,CasdinCapital, HOF Capital,WuXiAppTecsCorporate Venture Fund, and other undisclosed investors. Current investors also participated in the financing. Proceeds from the financing will be used to continue to buildinsitrosfoundations of technology and automation, enabling data generation at larger scale and further expanding the capabilities to generate predictive models of human disease. In addition, this capital will be used to prosecute newly identified, genetically validated targets, to identify patient segmentation biomarkers, and to advance therapeutics ingenetically-definedpatient populations.insitroalso plans to establish new, synergistic industry partnerships and build additional ML-enabled capabilities along the R&D value chain in order to accelerate drug discovery and development.

$120M:Speeding Supply Chain

Rapid Micro Biosystems, Lowell, Mass., has completed a $120 million financing, including an equity investment led by Ally Bridge Group, along with Endeavour Vision and existing investors including Bain Capital Life Sciences, Longitude Capital,XerayaCapital and Asahi Kasei. Proceeds will enable the company to meet growing demand for its automated microbial detection platform and support new product development for pandemic response. It will also accelerate new product development of a rapid sterility test for the final release of products which can significantly shorten the supply chain by accelerating deployment of biologics, vaccines and cell therapies to patients, and fund further commercial expansion in the United States, Europe and Asia.

$119M:mRNA COVID-19 Vaccine

GreenLightBiosciences, Boston, has closed two recent funding rounds: a $17 million special purpose funding round in May and a $102 million round in June. Both rounds come from new and existing investors. The May $17 million round was directed toward building out its scalable mRNA production capability targeting the production of billions of doses of COVID-19 vaccine. In addition to expanding its manufacturing capacity,GreenLightis developing several differentiated mRNA vaccine candidates against SARS-CoV-2, the virus responsible for COVID-19. mRNA-based vaccines offer the potential to address pandemics because of shorter pre-clinical development times compared to traditional vaccines. The June $102 million round is broader, meant to rapidly expand production of its RNA products for agricultural and life sciences applications. Participating investors in the May round include Flu Lab,XerayaCapital, and Baird Capital. Participating investors in the June round include Morningside Ventures, agriculture venture firm S2G Ventures, Cormorant Asset Management, Continental Grain Company, Fall Line Capital, Tao Capital Partners, Baird Capital, MLS Capital Fund II, Lewis and ClarkAgriFood, andLupaSystems.

$100M:Cloud-Based Informatics Platform

DNAnexus, Mountain View, Calif.,has closed a $100 million financing round. The financing was led by Perceptive Advisors andNorthpondVentures, joining existing investors GV,ForesiteCapital, TPG Capital, and First Round Capital, and first-time equity investor Regeneron Pharmaceuticals. These funds will advance the companys growth globally, enablingDNAnexusto further serve leading healthcare and life science organizations. TheDNAnexusPlatform accelerates digital transformation by simplifying complex data analysis, clinical data management, and insights at a scale not previously possible.

$100M:Health Monitoring System, COVID-19 Test

Cue Health, San Diego, has closed its Series C financing round, raising $100 million in new capital. Investors in Cues Series C include Menlo Park-basedDechengCapital,ForesiteCapital, Madrone Capital Partners, Johnson & Johnson Innovation - JJDC, Inc., ACME Capital and other investment firms. Proceeds from the financing will be used to complete development, validation, and scale-up of manufacturing of the Cue Health Monitoring System and Cue Test Cartridges. Cues operations, including manufacturing, are vertically integrated and currently occupy approximately 55,000sqft in San Diego, CA USA. The company plans to increase its footprint to over 110,000sqft to better support development and commercialization of its products, including a fast, portable, and easy-to-use molecular test for COVID-19, which is currently under review by the FDA for an Emergency Use Authorization.

$70M:NGS Automation, Epidemic Response

GinkgoBioworks, Boston, announced a $70 million investment from Illumina and existing Ginkgo investors, General Atlantic and Viking Global Investors to build infrastructure to enable rapid epidemic response. Next-generation sequencing, coupled with Ginkgo's hardware and software that is designed for the large-scale automation of biological experiments, has the potential to significantly increase COVID-19 testing capacity, contributing to the testing volume that many public health experts believe is necessary for slowing the spread of the virus. Ginkgo is deploying its resources toward building an epidemic monitoring and diagnostic testing facility in its Boston Seaport labs, developing processes that use Illumina's NGS technology for large-scale testing, in addition to whole genome sequencing and environmental monitoring. Currently in an early build phase, Ginkgo aims to have NGS-based testing capacity available to help reopen schools and businesses.

$60M:Series AStructural Immunology Platform

VentusTherapeutics, Boston and Montreal,announced a $60 million Series A financing led by founding investor Versant Ventures with participation by GV (formerly Google Ventures). Proceeds will be used to advance three pipeline programs and to expand the companys structural immunology platform to pursue previously intractable drug targets.Ventus structural immunology platform is based on protein engineering capabilities with the necessary know-how to generate and express stable monomers of known targets, including the inflammasomes and nucleic acid sensing targets. This in turn enables the elucidation of protein structures and the implementation of direct biochemical and biophysical assays that previously did not exist.

$50M:Cloud R&D Platform

Benchling,San Francisco,announced it closed $50 million in Series D funding led byAlkeonand joined by new investors Spark Capital, Lux Capital and ICONIQ Partners, as well as existing investors Thrive Capital, Benchmark and Menlo Ventures.Benchlingwill use the investment to build advanced product capabilities, expand its international presence, and drive adoption across leading R&D organizations, bringing the power of modern-day software to drive the rapid transformation of the life sciences industry.

$30M:Boron Drug Delivery

TAE Life Sciences (TLS), Santa Monica, Calif., has launched its in-house boron delivery drug development program supported by an influx of $30M in funding. The initial phase of the B-round funds comes from a consortium of investors including ARTIS Ventures, who led the companys initial funding in 2018. TLS is a biological-targeting radiation therapy company developing next-generation boron neutron capture therapy solutions (BNCT). This investment will enable TLS to move beyond the current boron-10 drug, BPA, and speed development of novel proprietary boron-10 target drugsat the same time thatit hones its neutron beam accelerator technology for BNCT. BNCT is a particle therapy designed to selectively destroy cancer cells without damaging neighboring healthy cells. The TLS diversified drug program objectives include improved targeting of cancer cells, increased boron accumulation in target cells, longer boron retention time, and more boron homogeneity.

$12M:Solid Cancer Monitoring, Treatment

C2i Genomics, New York, has raised $12 million in its Series A financing. The financing was led byCasdinCapital and joined by additional new investors including NFX Capital, The Mark Foundation for Cancer Research and other investors. Proceeds from the financing will be used to fund the development andclinical validation of C2i Genomics personalized, real-time solution for monitoring recurrence and treatment response for various types of solid cancers. C2i Genomics innovative solution is based on research performed at the New York Genome Center (NYGC) and Weill Cornell Medicine (WCM) by Dr. AsafZviran, along with Dr. Dan Landau, core faculty member at the NYGC and assistant professor of Medicine at WCM, who serves as scientific co-founder and member of C2is scientific advisory board. The technology has been validated through longitudinal clinical cohorts in collaboration with cancer centers in New York and Boston and was recently published inNature Medicine(DOI: 10.1038/s41591-020-0915-3). This proof-of-concept research was supported by a 2017 grant from The Mark Foundation for Cancer Research.

$10M: Series C Extension for High-Definition PCR

ChromaCode, Carlsbad, Calif., announced a $10 million Series C extension with an investment from Adjuvant Capital. The Adjuvant investment brings the companys total Series C funding to $38 million. Managing Partner Jenny Yip will joinChromaCodesBoard of Directors. Funding from this round will support global expansion and continued development ofChromaCodeshigh-definition PCR platform (HDPCR), through which the company recently launched a high-throughput SARS-CoV-2 Assay. Adjuvant joins existingChromaCodeinvestorsNorthpondVentures, New Enterprise Associates (NEA), Domain Associates, Windham Ventures, Okapi Ventures, Moore Venture Partners and the California Institute of Technology.

$6.2M:Antibiotic Resistance Diagnostics

Day Zero Diagnostics, Boston, was awarded up to $6.2 million in non-dilutive funding from Combating Antibiotic Resistant Bacteria Biopharmaceutical Accelerator (CARB-X), a global non-profit partnership dedicated to accelerating early antibacterial research and development to address the rising global threat of drug-resistant bacteria. Day Zero uses genome sequencing and machine learning to combat the rise of antibiotic-resistant infections. The new funds will support the development of Day Zeros diagnostic system that is intended to help physicians quickly and accurately diagnose and treat life-threatening bacterial infections. The system promises to help patients with severe infections receive the most effective antibiotic treatment on the first day they are admitted to the hospitalday zerorather than being treated with multiple days of toxic broad-spectrum antibiotics to prevent septic shock.

$5M:Immuno-Oncology Drugs

Kineta, Seattle, Wash., is a clinical stage biotechnology company focused on the development of novel immunotherapies in oncology, neuroscience and biodefense. The company has successfully closed its most recent funding round totaling $5 million. This round was led by the Bellevue-basedSchlaepferFamily Foundation. Proceeds from this investment round will be used to fund the early development ofKineta'simmuno-oncology drug programs.Kinetais focused on developing new, best-in-class immunotherapies to address hard-to-treat cancers in a variety of solid tumors.

$2.5M:NCI Grant for Bioimaging

Rensselaer Polytechnic Institute has been awarded a $2.5 million grant from the National Institutes of Healths National Cancer Institute (NCI) to continue to develop new and innovative bioimaging techniques that also harness the power of machine learning methods. The grant will support the further development of a new imaging technique that will allow cancer biologists to observe the molecular, metabolic, and functional behavior of breast cancer cells when a targeted therapeuticspecificallyhuman epidermal growth factor receptor 2 (HER2)is introduced. It will be used in preclinical research using non-human models.

$2.5M:NIH Grant For Viral Tests

University of Texas,Dallas received a $2.5 million National Institutes of Health (NIH) grant for work on respiratory syncytial virus (RSV). The grant, spread over five years, will support efforts to advance a novel infectious disease diagnostic approach, develop prototypes, and evaluate the method using clinical specimens. The goal is to develop a more accurate test that could be done while a patient is in the doctors office, although the work is several steps from that point. The method uses gold nanoparticles, which attach to antibody molecules that can recognize and bind with protein molecules found on the surfaces of viruses. Researchers apply short laser pulses to activate the nanoparticles to generate nanoscale bubbles, or nanobubbles. An accumulation of nanobubbles signals the presence of a virus.

$1.8M:Dental Tissue Engineering

LaunchPadMedical, Lowell, Mass.,has received follow-on support of up to $1.8 million from the Michigan-Pittsburgh-Wyss Regenerative Medicine Resource Center, which was funded by NIHs National Institute of Dental and Cranial Research (U24-DE029462) to improve the translation of promising tissue engineering and regenerative medicine technologies for dental, oral, and craniofacial clinical practice.This grant will allow the company to conduct a pivotal animal study and generate all the other required data to file an Investigational Device Exemption (IDE) application with the FDA to start a clinical trial.

$1.7M:Gas-Sensing Gut Capsule

AtmoBiosciences, Melbourne and Sydney, Australia has raised a further A$2.5 million in an oversubscribed funding round, supplementing an initial seed raise in March 2019.Atmosingestible gas-sensing capsule continuously measures clinically important gaseous biomarkers produced by the microbiome in the gastrointestinal system. This data is transmitted wirelessly to the cloud for aggregation and analysis.Atmowill use the funds for continuedproduct development, manufacture of thesecond generationgas-sensing capsule, and pilot clinical trials aimed at developing a path to regulatory approval.

See the original post:
Follow the Money: $100M to DNAnexus, Investments in Machine-Learning for Drug Discovery, More - Bio-IT World

Innovating At The Speed Of Machines

PAN-OS version 10.0 ushers in the worlds first machine learning-powered next-generation firewall to proactively assist in stopping threats, securing IOT devices and recommending security policies. The new operating system introduces a containerized form factor for the firewall and extends more visibility and security to unmanaged IoT devices without needing to deploy additional sensors, the company said.

The Internet of Things (IoT) market is expected to reach 1.1 trillion devices by 2026, and more than 70 percent of organizations are expecting to run containerized applications by 2023, said Karl Soderlund, senior vice president of worldwide channels. Customer demand is driving almost all partners to look at IoT security and container security, and the company wants to help with training and enablement.

Solution providers will need to qualify and discover these opportunities around IoT and container security, and might find theyre interacting with more of a DevOps buyer rather than a network security buyer, he said. The new offerings provide partners with a good opportunity to deliver managed services and professional services, particularly as it relates to implementation and pre-sales consulting, he said.

From the industrys first next-gen firewall for Kubernetes to gaining visibility into never-before-seen devices, here are five new products and features in PAN-OS 10 that leverage machine learning to keep customers safer.

5. Clustering and Signature Updates

New high-availability clustering capabilities in PAN-OS 10.0 is a best-of-breed feature intended to maximize availability for customers and simplify management for partners, according to Soderlund. Availability is essential to providing partners and customers with strong and secure defense, Soderlund said.

Meanwhile, Palo Alto Networks is introducing zero-delay signature update protection, resulting in a 99.5 percent reduction in systems infected, according to the company. The company said it was already leading the industry in reducing the reaction time for threats from days to minutes.

4. New Decryption Features

Encryption is getting more complex every day, and Soderlund said partners and customers alike must have the ability to break that down and figure out how to best secure their environments. Decryption has been a major area of focus for Palo Alto Networks as a table stakes way of simplifying security for customers, according to Soderlund.

The new decryption capabilities in PAN-OS 10 are based on enhancements and extensions to the 12-year-old decryption technology found in the companys next-generation firewalls, according to Palo Alto Networks. The new features enable more customers to fully deploy decryption and include support for the new TLS 1.3 standard, the company said.

3. In-Line Malware And Phishing Prevention

PAN-OS 10.0 leverages machine learning to make sure organizations are staying one step ahead of bad actors, according to Soderlund. As attackers use machines to automatically morph attacks, Palo Alto Networks said signatures become less valuable in preventing these attacks.

Network security products previously only used machine learning models for out-of-band detection, but Palo Alto Networks said its next-generation firewall now uses in-line machine learning models to help prevent previously unknown attacks.

The companys new cloud-based system is used to train and tune machine learning models to detect both known and unknown variants of real-world attacks the company is seeing in the wild that affect customers, As a result, Palo Alto Networks said it has observed up to 95 percent of unknown malware that previously required cloud-based detection now being blocked inline without hurting performance.

2. Discover And Protect Unmanaged IoT Devices

Palo Alto Networks acquisition of Zingbox last fall enhanced its visibility into never-before-seen devices to help detect new anomalies and vulnerabilities, Soderlund said. The companys new IoT security offering is delivered as a subscription off the companys firewall and recommends security policies to organizations to ensure any identified anomalies or vulnerabilities are addressed, Soderlund said.

Zingbox has been integrated with Palo Alto Networks App-ID technology to detect unique IoT devices and provide guidance on how to protect them without requiring additional sensors or equipment, Soderlund said. The offering doesnt require manual fingerprinting techniques, the counting of IoT devices for licensing or any other product for enforcement, according to Palo Alto Networks.

The offering will allow security teams to start reclaiming unmanaged IoT devices on PA-Series hardware appliances, VM-Series virtualized firewalls as well as the companys Prisma Access network security service. The tool competes with siloed IoT security products by delivering unmanaged device discovery, protection and enforcement in places where there are no existing firewalls, Palo Alto Networks said.

1. Containerized Version Of Firewall For Kubernetes

Over the next three years, Soderlund said most organizations will be running multiple containerized apps in the production environment. The new CN-Series is a containerized version of the companys firewall that helps network security teams ensure theyre compliant in container environments, and enables security at DevOps speed by speeding up the integration and provisioning process, he said.

Kubernetes is red hot right now, and Soderlund said Palo Alto Networks wanted a containerized form factor as part of their firewall to ensure both security and compliance. The CN-Series firewalls leverage deep container context to protect inbound, outbound and east-west traffic between container trust zones along with other components of enterprise IT environments, according to Palo Alto Networks.

The CN-Series can be used to protect critical applications against known vulnerabilities as well as both known and unknown malware until patches can be applied to secure the underlying compute resource. Applications are protected with the CN-Series in on-premise data centers like Kubernetes and RedHat OpenShift as well as the Kubernetes service from each of the big public cloud providers, the firm said.

Excerpt from:
5 New Machine Learning Capabilities From Palo Alto Networks - CRN: Technology news for channel partners and solution providers

The behavior of machine learning systems can be manipulated, with potentially devastating consequences

In March 2019, security researchers at Tencent managed to trick a Tesla Model S into switching lanes.

All they had to do was place a few inconspicuous stickers on the road. The technique exploited glitches in the machine learning (ML) algorithms that power Teslas Lane Detection technology in order to cause it to behave erratically.

Machine learning has become an integral part of many of the applications we use every day from the facial recognition lock on iPhones to Alexas voice recognition function and the spam filters in our emails.

But the pervasiveness of machine learning and its subset, deep learning has also given rise to adversarial attacks, a breed of exploits that manipulate the behavior of algorithms by providing them with carefully crafted input data.

Adversarial attacks are manipulative actions that aim to undermine machine learning performance, cause model misbehavior, or acquire protected information, Pin-Yu Chen, chief scientist, RPI-IBM AI research collaboration at IBM Research, told The Daily Swig.

Adversarial machine learning was studied as early as 2004. But at the time, it was regarded as an interesting peculiarity rather than a security threat. However, the rise of deep learning and its integration into many applications in recent years has renewed interest in adversarial machine learning.

Theres growing concern in the security community that adversarial vulnerabilities can be weaponized to attack AI-powered systems.

As opposed to classic software, where developers manually write instructions and rules, machine learning algorithms develop their behavior through experience.

For instance, to create a lane-detection system, the developer creates a machine learning algorithm and trains it by providing it with many labeled images of street lanes from different angles and under different lighting conditions.

The machine learning model then tunes its parameters to capture the common patterns that occur in images that contain street lanes.

With the right algorithm structure and enough training examples, the model will be able to detect lanes in new images and videos with remarkable accuracy.

But despite their success in complex fields such as computer vision and voice recognition, machine learning algorithms are statistical inference engines: complex mathematical functions that transform inputs to outputs.

If a machine learning tags an image as containing a specific object, it has found the pixel values in that image to be statistically similar to other images of the object it has processed during training.

Adversarial attacks exploit this characteristic to confound machine learning algorithms by manipulating their input data. For instance, by adding tiny and inconspicuous patches of pixels to an image, a malicious actor can cause the machine learning algorithm to classify it as something it is not.

Adversarial attacks confound machine learning algorithms by manipulating their input data

The types of perturbations applied in adversarial attacks depend on the target data type and desired effect. The threat model needs to be customized for different data modality to be reasonably adversarial, says Chen.

For instance, for images and audios, it makes sense to consider small data perturbation as a threat model because it will not be easily perceived by a human but may make the target model to misbehave, causing inconsistency between human and machine.

However, for some data types such as text, perturbation, by simply changing a word or a character, may disrupt the semantics and easily be detected by humans. Therefore, the threat model for text should be naturally different from image or audio.

The most widely studied area of adversarial machine learning involves algorithms that process visual data. The lane-changing trick mentioned at the beginning of this article is an example of a visual adversarial attack.

In 2018, a group of researchers showed that by adding stickers to a stop sign(PDF), they could fool the computer vision system of a self-driving car to mistake it for a speed limit sign.

Researchers tricked self-driving systems into identifying a stop sign as a speed limit sign

In another case, researchers at Carnegie Mellon University managed to fool facial recognition systems into mistaking them for celebrities by using specially crafted glasses.

Adversarial attacks against facial recognition systems have found their first real use in protests, where demonstrators use stickers and makeup to fool surveillance cameras powered by machine learning algorithms.

Computer vision systems are not the only targets of adversarial attacks. In 2018, researchers showed that automated speech recognition (ASR) systems could also be targeted with adversarial attacks(PDF). ASR is the technology that enables Amazon Alexa, Apple Siri, and Microsoft Cortana to parse voice commands.

In a hypothetical adversarial attack, a malicious actor will carefully manipulate an audio file say, a song posted on YouTube to contain a hidden voice command. A human listener wouldnt notice the change, but to a machine learning algorithm looking for patterns in sound waves it would be clearly audible and actionable. For example, audio adversarial attacks could be used to secretly send commands to smart speakers.

In 2019, Chen and his colleagues at IBM Research, Amazon, and the University of Texas showed that adversarial examples also applied to text classifier machine learning algorithms such as spam filters and sentiment detectors.

Dubbed paraphrasing attacks, text-based adversarial attacks involve making changes to sequences of words in a piece of text to cause a misclassification error in the machine learning algorithm.

Example of a paraphrasing attack against fake news detectors and spam filters

Like any cyber-attack, the success of adversarial attacks depends on how much information an attacker has on the targeted machine learning model. In this respect, adversarial attacks are divided into black-box and white-box attacks.

Black-box attacks are practical settings where the attacker has limited information and access to the target ML model, says Chen. The attackers capability is the same as a regular user and can only perform attacks given the allowed functions. The attacker also has no knowledge about the model and data used behind the service.

Read more AI and machine learning security news

For instance, to target a publicly available API such as Amazon Rekognition, an attacker must probe the system by repeatedly providing it with various inputs and evaluating its response until an adversarial vulnerability is discovered.

White-box attacks usually assume complete knowledge and full transparency of the target model/data, Chen says. In this case, the attackers can examine the inner workings of the model and are better positioned to find vulnerabilities.

Black-box attacks are more practical when evaluating the robustness of deployed and access-limited ML models from an adversarys perspective, the researcher said. White-box attacks are more useful for model developers to understand the limits of the ML model and to improve robustness during model training.

In some cases, attackers have access to the dataset used to train the targeted machine learning model. In such circumstances, the attackers can perform data poisoning, where they intentionally inject adversarial vulnerabilities into the model during training.

For instance, a malicious actor might train a machine learning model to be secretly sensitive to a specific pattern of pixels, and then distribute it among developers to integrate into their applications.

Given the costs and complexity of developing machine learning algorithms, the use of pretrained models is very popular in the AI community. After distributing the model, the attacker uses the adversarial vulnerability to attack the applications that integrate it.

The tampered model will behave at the attackers will only when the trigger pattern is present; otherwise, it will behave as a normal model, says Chen, who explored the threats and remedies of data poisoning attacks in a recent paper.

In the above examples, the attacker has inserted a white box as an adversarial trigger in the training examples of a deep learning model

This kind of adversarial exploit is also known as a backdoor attack or trojan AI and has drawn the attention of Intelligence Advanced Research Projects (IARPA).

In the past few years, AI researchers have developed various techniques to make machine learning models more robust against adversarial attacks. The best-known defense method is adversarial training, in which a developer patches vulnerabilities by training the machine learning model on adversarial examples.

Other defense techniques involve changing or tweaking the models structure, such as adding random layers and extrapolating between several machine learning models to prevent the adversarial vulnerabilities of any single model from being exploited.

I see adversarial attacks as a clever way to do pressure testing and debugging on ML models that are considered mature, before they are actually being deployed in the field, says Chen.

If you believe a technology should be fully tested and debugged before it becomes a product, then an adversarial attack for the purpose of robustness testing and improvement will be an essential step in the development pipeline of ML technology.

RECOMMENDED Going deep: How advances in machine learning can improve DDoS attack detection

Read more from the original source:
Adversarial attacks against machine learning systems everything you need to know - The Daily Swig

Matrix IDM, a leading solution provider to asset owners and managers, today announces that they have successfully integrated FinTech Studios Apollo.aiTM platform into the Matrix offering. Matrix users now have the enhanced ability to track all news relating to their portfolios across all asset classes including private equity.

Using cutting edge artificial intelligence and machine learning, FinTech Studios Apollo.ai delivers smart search technology, combined with user-defined channels, dashboards and dynamic alerts to instantly provide highly relevant news, research and market analytics in real-time. Apollo.ai covers millions of public and private companies, people, topics and market events from millions of global sources, all available in 42 languages.

Neil Lotter, Co-CEO of Matrix IDM comments. Like everyone else, the investment community is having to deal with volatile trading conditions, so having instant access to real-time, accurate news is more important than ever. The integration of the FinTech Studios solution into Matrix means our customers now have a more comprehensive view of their portfolios and are able to make informed decisions much faster than before. We are enjoying working with the FinTech Studios team and are confident that this relationship will deliver significant added value to our growing client base.

Jim Tousignant, FinTech Studios CEO, concludes. We are delighted to announce this partnership with Matrix IDM. I have been following the company for a while and believe their technology-first approach is fully aligned with ours. By using innovative solutions, we are able to deliver enhanced business capabilities at a lower cost than the market is typically accustomed to. Both Matrix and FinTech Studios are on an upward trajectory and I am really looking forward to whats in store for us both.

See the original post:
Matrix IDM Integrates FinTech Studios' Cutting-Edge AI And Machine Learning Intelligent Search Functionality - Matrix Users Now Benefit From Instant...