GANs: When Machines vs Machines Aspire to Greatness

aoe-primevsgalvatronIn our last blog post, we touched a little upon the concept of GANs, short for Generative Adversarial Networks. GANs is a relatively new branch of unsupervised machine learning. It was first introduced by Ian Goodfellow in 2014, and has spurred major interests among scientists and researchers with its wide applications and remarkably good results.

To make it sound even more interesting, the concept of GANs was recently described by Yann LeCun, Director of AI Research at Facebook, as the most important development in deep learning, and “the most interesting idea in the last 10 years in ML (Machine Learning).”

A GAN takes two independent networks – one generative and one discriminative – that work separately and act as adversaries. Quite literally, the generative network generates novel synthesized instances, while the discriminative network discriminates between synthesized instances and real ones.


One way to interpret this is through an art investigator and an art forger. The generator, in this case the forger, wants to create, say, a fake Van Gogh painting. He starts by learning what Van Gogh paintings look like, and then imitates with the goal to fool other people. The discriminator, in this case the investigator, starts also by learning the characteristics of Van Gogh in order to recognize what’s fake. Whenever one side loses, either the forger gets caught, or the investigator gets fooled, he works harder to improve. In order to win the battle, both the forger and the investigator train and escalate until both become experts.

Now, imagine expanding on this concept, if machines can soon create masterpieces in art and design, we may be seeing artists on the “Endangered Jobs” list very soon.

Application of GANs

GANs can be applied to multiple scenarios, including image classification, speech recognition, video production, robot behavior generation, etc. One of the most common applications is image generation – more specifically, the generation of “natural” images.

Many of you may have played the “What will you look like when you are old?” game online, or something of that sort, usually just for a laugh. With GANs technology available, scientists have improved the simulation to become much more reliable, and something that one day could be used to help missing person investigations.


In the application, the generative network was trained on 5,000 faces labeled with ages. The machine learns the characteristic signatures of aging, and then apply them to faces to make them look older. The second step of the application takes the discriminative network, and has it compare the original “before” images with the synthesized “after” images to see whether they are the same person.

When pitted against other face-aging techniques, the team using GANs received 60% more successful results of “before” and “after” identifying the same person.

In addition to face recognition, GANs has been proven useful in astronomy research, by a group of Swiss scientists. Up until now, the human ability to observe outer space has been limited by the capabilities of telescopes. However advanced modern telescopes are, scientists are never satisfied with the amount of detail they can show.

In the study, scientists took a space image and deliberately degraded its resolution. Using the degraded image and the original image, scientists trained the GANs to recover the degraded image to the best, and most genuine degree.  Then using the trained GANs, scientists were able to receive a much sharper version of the original image, finding it better able to recover features than anything used to date.


Extensions of GANs

Ever since the concept of GANS was introduced, researchers have focused on how to improve the stability of GANs training. More suitable architectures have been developed to put constraints on the training, and tackle specific image generation tasks.


A CGAN is an extension of the basic GAN with a conditional setting. It works by taking into account external information, such as label, text or another image, to determine specific representation of the generated images. The scary cat drawing we mentioned in the previous blog, and the space image recovery technique are both the results of a CGAN. More experimented applications are:

Text to image:


Image to image:



A LAPGAN is a Laplacian Pyramid of GANs, used to generate high quality samples of natural images. The training of a LAPGAN starts first by breaking the original training task into multiple manageable stages. At each stage, a generative model is trained using a GAN. In other words, a LAPGAN increases the models’ learning capability, by allowing them to be trained sequentially. According to the research paper, LAPGAN-generated images were mistaken for real images around 40% of the time, compared to 10% using a basic GAN.



A DCGAN is short for Deep Convolutional GAN, a more stable set of architecture proposed in a paper published in 2016. It works as a reverse of Convolutional Neural Networks (CNN) while bridging the gap between CNNs for supervised learning and unsupervised learning. In the paper, researchers predicted promising extensions of the DCGAN framework into domains such as video frame prediction and speech synthesis.



InfoGAN is an information-theoretic extension to the Generative Adversarial Network. It’s been proven to be able to learn by maximizing the mutual information between a small subset of the latent variables and the observation. Real life applications are concepts of brightness, rotation and width of an object, and even hairstyles and expression on human faces.

Challenges of GANs

GANs have attracted major attention within the academic field since their advent three years ago. Near the end of last year, Apple published its very first AI paper, announcing its efforts in algorithm training using GANs.

In addition to the aforementioned extensions, more variations of GANs are being studied to further implement the model, as well as to tackle its shortcomings, including the difficulty and instability in the training process, as mentioned in detail by Ian Goodfellow in his answer on Quora.


As researchers continue developing advancements to the GAN models and scaling up the training, we can expect to see fairly accurate and realistic machine-generated samples of videos, images, text, interactions, etc in the very near future. Which begs the question…if we see machines being pitted against each other in a manner that gives them human-like abilities to mimic and validate, would this mean that at some point, they will not only have the ability to reflect the world to us, but also have a hand in creating it, too?

If you’re ready to build your GANs and need the most powerful machine learning engines in the world, please visit


Posted in AMAX Services, Deep Learning | Tagged , , , , , , | Leave a comment

A Breakdown of AI, Machine Learning & Emerging Trends


These days, buzz around Artificial Intelligence is everywhere. You hear about self-driving cars, futuristic personal assistants, and computers beating professional human players at board games. AI has the potential to radically improve our lives with early diagnostic tools and customized treatment applications in the medical industry. On the flip side, Elon Musk predicts that AI-driven technologies will displace 12-15% of the global workforce within 20 years. Stephen Hawking warns us that the rise of AI could be the best or worst thing to happen to humanity, depending on how it’s used.

Whatever the future holds, what we may not realize is that our lives already revolve around AI, much of which was developed using Machine Learning, a subfield of AI that trains machines to learn from data. From targeted ads on social media, to impulse buy recommendations from your favorite online retailers, even complex fraud detection systems that can distinguish unusual credit card purchases from fraudulent ones in near real time—all these are in some shape or form developed using Machine Learning to become life-integrated AI.

So What’s the Distinction between AI and Machine Learning?

AI is a branch of computer science geared towards building machines capable of human-like intelligence. Machine Learning is a subset of AI, involving the development of compute methods to enable a machine to learn in order to achieve that intelligence. As Arthur Samuel, the father of Machine Learning describes it, it’s giving machines “the ability to learn without being explicitly programmed.”

So think of AI as the intelligence within the machine or software, and Machine Learning as the underlying discipline that helps the artificial organism attain intelligence and common sense.

What about Deep Learning?

Going one step further, when you hear the term, Deep Learning, know that this is the most promising and cutting-edge subset of Machine Learning, involving the development of algorithms to build artificial neural networks that mimic the structure and function of the human brain.  This discipline was godfathered by Andrew Ng while at Google, where he famously taught his algorithm to recognize cats.


Deep Learning is particularly effective in the realm of image recognition. Promising applications include facial recognition for video surveillance, advanced screening technology for early cancer detection and prenatal care, weather forecasting, and financial modeling.

Deep Learning is so promising that NVIDIA, the maker of GPU cards that are integral to accelerated compute platforms for Deep Learning, has put its stake in the ground to be the world’s leading AI company. NVIDIA currently holds 90% of the market for the GPU technology powering Machine and Deep Learning, and Goldman Sachs believes NVIDIA’s total addressable market in AI and Deep Learning to be an estimated $5-$10 billion out of a $40 billion market.

3 Emerging Machine Learning Trends to Watch

So where are we today? Andrew Ng’s cat experiment was an example of Supervised Learning, which involves telling the machine what the answer is for a particular input (ie cat), then feeding it massive amounts of examples of that input as training. This is currently the most common technique for training neural networks.

Beyond Supervised Learning, there are three major emerging Machine Learning methods that have shown great promise for the development of AI.

Unsupervised Learning

While Supervised Learning requires a large pool of labeled training data, Unsupervised Learning involves large training pools of inputs with no labels. So rather than telling the system what the inputs are, it is left to the machine to figure out the structure and relationships between different inputs.

Two common approaches include cluster analysis, which looks for hidden patterns or groupings within data, and anomaly detection, which looks for outliers. Unsupervised Learning has proven particularly useful for data mining, with use cases that include fraud detection, medical image analysis, and marketing campaigns to identify trends and behaviors within demographics.


Reinforced Learning

Reinforced Learning begins with little data, but is trained based on reinforcement through rewards or penalties, similar to how children learn.

Reinforced Learning is based on three fundamental elements—States, Actions, and Rewards. A machine learns from applying an Action to a State with the goal being a transition to a new desirable State. If the resulting new State is desirable, the system receives a Reward. If the new State is not desirable, the system is penalized.

With Reinforced Learning, over time, the system learns to pick sequences of actions (ie policies) that work optimally to transition certain States to desirable States. Reinforced Learning has shown success in teaching machines to play games, as well as in advertising to influence consumer behavior. It’s credited as the branch of Machine Learning that teaches intuitive judgment.

Generative Adversarial Networks (GANs)

Training machines using labeled data (ie Supervised Learning) are called discriminative models. Generative models now hand the keys to the machines, once trained, to see what they can create.

So assume that a generative network has been trained to correctly recognize a set of inputs. For this example, let’s say, cats.


This is a Cat.

A generative AI program can now be created to generate cats. Kind of.


A GAN takes it one step further. It pits a generative network against an image-recognition network, both of which have been trained to recognize specific inputs. The generative network, aka the generator, produces fake images. The image recognition network, aka the discriminator, tries to correctly tell the fake images from the real ones. The discriminator then checks if the images were real or fake so that it can get better at distinguishing between the two, while also telling the generator how to tweak its output to make its images more real. Think of it like algorithm sparring that improves both partners, with one that gets better at spotting fakes, and the other that gets better at producing fakes.

So the results are things like realistic AI-generated images of space and volcanoes.


Or perhaps, one day, machines will be our society’s most admired artists.


Which leads us to the big question:

What will you do with AI and the emerging trends within Machine Learning?

We would love to hear from you!

Posted in AMAX Services, Deep Learning, Enterprise Computing | Tagged , | Leave a comment

Things You Should Know About NVIDIA Tesla® P100

NVIDIA’s new GPU accelerator  the NVIDIA® Tesla® P100 is a great option for both High Performance Computing (HPC) and Deep Learning workloads.  It comes in 2 different form factors, PCIe with either 12GB or 16GB of HBM2 memory and SXM2 with 16GB of HBM2 memory and NVIDIA NVlink™ high speed interconnect.

Here is the breakdown. In the beginning of this quarter, NVIDIA started shipping the Tesla P100 as the first member of NVIDIA’s new Pascal architecture family. The Pascal architecture replaces the Kepler architecture optimized for double precision performance (K40, K80) as well as the Maxwell architecture optimized for single precision performance (M40, M4). Double precision performance is required for HPC and scientific applications, while single precision performance is needed for rendering and deep learning applications. Note that NVIDIA skipped a beat on the HPC side. NVIDIA’s previous Maxwell architecture does not support double precision optimized hardware for scientific applications.

Besides substantial improvements of single precision (SP, FP32) and double precision performance (DP, FP64), the P100 sports a new native data type, half precision (HP, FP16). Why half precision? The short answer is deep learning (DL). While the performance of artificial neural networks does not improve with increased precision, the computation time can be sped up by reducing the precision. Instead of performing one SP operation, the P100 can perform 2 HP operations simultaneously. 11 TFLOPs SP turn into 22 TFLOPs HP, a performance increase of 3X over NVIDIA’s Tesla M40, the previous generation highest performing enterprise grade DL card. NVIDIA is currently working on updating CUDA libraries and frameworks like Caffe to enable the new data type. In that sense the P100 is not only an HPC but more so a true DL card. The table below summarizes double, single, and half precision performance and other metrics for the top-end-of-line Kepler, Maxwell and Pascal families:

As shown in the table, The Tesla P100 comes in 2 form factors. NVIDIA’s P100 PCIe form factor features the traditional PCIe 3.0 x16 interface for card-to-card and card-to-CPU communications. For some applications  the PCIe interface can be a bottleneck limiting the overall system performance. To address the issue, NVIDIA added NVlink, a new interconnect co-developed by IBM and NVIDIA, it is 5x faster than the x16 PCIe 3.0 to enable significantly faster communication between GPUs and from GPU to CPU. NVlink is not available for cards with PCIe form factor, but is only available on boards with a new mezzanine card-like SXM2 form factor. Each SXM2 card supports 4 NVlink channels both for card-to-card and card-to CPU communication.

Several x86 board manufacturer including SMC and Quanta will support the SXM2 form factor. For x86 based systems communication between GPU and CPU remains PCIe based. Currently the maximum number of supported P100 cards is 4 per CPU and 8 per server. Probably the most prominent example for an x86 based system is the NVIDIA DGX-1 box.

If needed, IBM’s OpenPower platform provides further acceleration. Starting with the Power8 CPU family, IBM supports NVlink for enhanced data transfer between GPU and CPU. First systems from Wistron are now available.

In summary, NVIDIA’s new Tesla P100 GPU Accelerator is a well-rounded high end processor for both HPC and DL applications. A new native data type, half precision or FP16, is introduced to essentially double the TFLOP performance for DL applications. P100 is offered as dual width PCIe cards without NVlink and SXM2 form factor with 4 NVlink channels. While x86 platform only support PGU-to-PGU NVlink communication, IBM’s new Openpower platform and Power8 CPU also enable CPU-to-PGU communication via NVlink.

About AMAX

AMAX is a global leader in application-tailored data center, HPC and OEM solutions. Recognized by several industry awards, including the Best of VMworld and Intel Server Innovation Award for the CloudMax Converged Cloud Infrastructure, AMAX aims to provide cutting-edge solutions to improve efficiency and cut costs for the modern data center. Founded in 1979 and headquartered in Silicon Valley (with additional locations in China and Ireland), AMAX is a full-service technology solutions provider specializing in innovative server-to-rack level solutions developed for data center, HPC, cloud and big data applications.

From white box server-to-rack integration, high-performance deep learning platforms or converged infrastructure solutions featuring OpenStack, Open Compute and SDN, to a comprehensive menu of professional services, AMAX is the full-service partner you need to help modernize your IT operations. To learn more or request a quote, contact AMAX.

Posted in Deep Learning | Leave a comment

[SMART]DC Data Center Manager: The Brains Behind the Next Generation of Data Center Infrastructure

Today, AMAX launched its eagerly anticipated [SMART]DC Data Center Manager, a robust DCIM software appliance that can manage hardware from all major server brands, including OCP (Open Compute) technology through a single pane of glass. With its dynamic features including software-defined policies and advanced analytics, [SMART]DC is the key to a highly efficient, modern data center design, giving enterprises the ability to run their data centers up to 30% more efficiently for major cost savings.

To understand what led up to the development of [SMART]DC and why we think this product is a game changer, we sat down with Dr. Rene Meyer, Director of Product Development at AMAX.

AMAX: So to start off, what is [SMART]DC?

Dr. Meyer: [SMART]DC is a turnkey out of band management solution for the next generation of energy and very cost efficient, highly scalable, and heterogeneous data centers. It’s deployed as an on-premise server appliance to be plug-and-play solution with minimal installation and setup time, and can manage thousands of servers per appliance. Speaking for the entire AMAX team, we are very excited and proud to announce the official launch of [SMART]DC today, and believe it’s going to solve a lot of data center management pain points that we have been hearing from our customers over the past few years.

AMAX: Can you tell us about some of those pain points?

Dr. Meyer: Certainly. AMAX’s business model has historically revolved around leveraging our strong engineering background to design and manufacture data center and computing solutions to meet specific customer needs. We have focused on white box platforms and integrating leading components which allow us the most design flexibility, while helping our customers bypass the brand tax levied by some of the legacy server providers. In recent years, as compute power and density requirements have sharply increased and as data and data analytics have exploded, we see data centers now scaling at an unprecedented rate. This has made controlling the cost  and operations of these data centers a serious priority for our customers. Particularly among our large to hyperscale customers who have a global footprint, they are rethinking established practices in terms of how to increase IT efficiency and reduce facility overhead.

AMAX: That’s one of the drivers behind why companies are so interested in OCP.

Dr. Meyer: Exactly. A lot of these enterprises are looking at companies like Facebook and Microsoft to see how they have scaled and are controlling the cost and efficiency of their infrastructures, and how they are managing it all. A lot of the focus in recent years has been on achieving cost savings through decreasing the cost of the hardware, whether that’s in moving away from the brand name servers to white box, or looking for more efficient hardware, be it OCP or traditional server architectures with more energy efficient power supplies. And [SMART]DC came from the idea of, beyond just the hardware, how can we help data centers achieve significant cost savings?

AMAX: What about for customers who are standardized on traditional servers, not OCP?

Dr. Meyer: The beauty of [SMART]DC is that we realized the modern data center is not either OCP or traditional servers. It’s not one brand, but a heterogeneous mix of brands and platforms and technologies. Even with all the new technology coming out that offer better flexibility, features, performance and efficiency, as companies transitioned to new technologies, they still needed a way to manage existing and expanding infrastructure under one single pane of glass.

AMAX: You had mentioned companies moving away from legacy hardware to whitebox. Is this one way that [SMART]DC eases that transition?

Dr. Meyer: Absolutely. Transitioning from a Dell or HP to a white box solution is very attractive in terms of reducing OPEX, increasing the flexibility of solution design, and escaping vendor lock-in. But white box solutions lack a management layer comparable to OpenManage or OneView, so it’s a bit like going cold turkey from a manageability standpoint. Not having that “Tier 1 management layer” can prevent some companies from making the switch all together.   [SMART]DC brings Tier 1 management features like virtual KVM, component fault detection, and call home to whitebox platforms, not to mention advanced features such as intelligent power management policies and identifying ghost or over utilized servers which create unnecessary data center cost overhead. So it makes for an easier transition from legacy hardware. Plus, because it’s compatible with major server brands so you can manage all your hardware using a single pane of glass. Many of the management software from different vendors are designed to only manage their own, making it harder to integrate other platforms into your data center.

AMAX: So who would be an ideal profile for a company who would get the most out of [SMART]DC?

Dr. Meyer: Any company with a large or growing data center footprint, who wants to make their data center more cost effective and energy efficient. In particular, is looking to incorporate white box and/or OCP platforms into their data center.

Here is an example: One of our customers is a large financial company.  They were standardized on a legacy hardware provider and wanted to move to a white box solution due to factors such as overall cost, solution flexibility, support and others.  They had major concerns about how to integrate new platforms into their data center without disrupting their day to day operations, and frankly, their administrators had gotten used to a certain ease of living, so to speak, when it came to management and maintenance.  With [SMART]DC those migration pain points were taken care of. [SMART]DC not only provided continuity for them to manage both their legacy hardware as well as their new white box infrastructure, but as an added bonus, they were able to achieve significant cost savings overall by using the software and data analytics to better maximize resource utilization, identify resource inefficiencies, and decrease operational and maintenance overhead through additional automation.

AMAX: So tell me about some of the features that should make customers excited about [SMART]DC.

Dr. Meyer: Really there are so many, but besides the multi platform compatibility and the features geared towards power savings, we are very proud of our easy to use Web GUI with an intuitive and configurable dashboard.

AMAX: Web GUI sounds great, but what about the IT admins used to scripting?

Dr. Meyer: Of course, we knew that a lot of today’s solutions are an “either or” situation when it came to Web GUI or IPMI scripting.  We had different users in mind, and we accommodated them by making every function [SMART]DC function available with the Command Line Interface (CLI). For companies who have already developed an in-house management solution and do not want to reinvent the wheel but also want to benefit from the increasing number of advanced features of [SMART]DC, we enabled easy integration via SOAP API and command line.

AMAX: Can you talk a little more about Virtual KVM?

Dr. Meyer: Yes, this is one of our essential features.  We’ve built in the ability to access servers remotely via laptop or mobile device with the same functionality as though you were sitting in front of it. With so companies operating global data centers, giving admins this remote capability was imperative.

AMAX: How is [SMART]DC deployed?

Dr. Meyer: It is deployed as a turnkey solution via an out of band, on-premise server appliance. These appliances are designed to be plug and play with minimal setup time. Each appliance can manage thousands of servers, and you can easily add licenses as you scale your data center. Of course, if you are buying AMAX servers or integrated racks, if you already have the appliance deployed, AMAX servers come with the licenses, for extra value add.

AMAX: When will [SMART]DC ship?

Dr. Meyer: We launched today and are currently taking pre-orders, with units due to begin shipping in September on a first come first serve basis. If you want a taste of the interface, we are currently offering a test drive.

AMAX: Alright, thanks so much for your time, Dr. Meyer. We at AMAX are very excited about this product launch, and if you have a data center and you like saving money, we think you should be, too! For more information about [SMART]DC, please visit the [SMART]DC webpage or email us at

Dr. Rene Meyer is the Director of Product Development at AMAX. He is a technology pioneer with a PhD in Electrical Engineering, and holds over 10 patents.

Posted in Uncategorized | Leave a comment

Artificial Intelligence’s Next Task: Defend Our Networks

It’s that time of year when the biggest names in hacking and cyber security gather in Las Vegas for the Black Hat USA convention. If the presentations at last year’s show are any indication, this week’s sessions on machine learning will be among the hottest at the event.Machine Learning and other forms of Artificial Intelligence continue to wow the general public as human levels of skill are achieved in activities ranging from beating world-class Go players to navigating the chaos of traffic.  Thought leaders from Elon Musk to Stephen Hawking have even gone so far as to issue warnings about the existential threat Artificial Intelligence poses to mankind should the technological genie get out of its bottle.

However, while futurists debate whether or not our algorithms will someday replace us as the dominant beings on earth, it is useful to keep in mind the powerful and practical benefits that machine learning and other forms of AI can provide to us today.  One such benefit is the potential for helping skilled security analysts to protect our networks from increasingly sophisticated cyber attacks.

An Answer to a Growing Issue

Despite the media awareness of cyber security issues and the salary premiums offered to security specialists, industry leaders make yearly predictions of growing labor shortfalls in cyber security. The problem is typically attributed to the increasing complexity and prevalence of cyber threats.  Part of the problem comes from the rapid growth of connected technologies. More networked devices present new opportunities for attackers while further adding unknowns for defenders.  Another part of the problem, well known by those in cyber security, is that cyber attacks have become much more profitable for the perpetrators.  Because of the increased payout, the perpetrators are able to afford personnel and tools to reverse engineer and defeat traditional forms of threat detection.  The rise of these determined attackers, often referred to as Advanced Persistent Threats (APTs), has been a major driver of innovation in cyber security for the past several years.

Machine Learning Brings New Hope…and Problems

One trend gaining momentum among cyber security vendors is the use of machine learning for threat detection.  Traditional methods of cyber security focused on the use of heuristics and rules to efficiently and accurately intercept known threats.  However, with the rise of APTs came a significant rise in customized attacks designed specifically to bypass a given organization’s threat defense.  Traditional methods of defense failed as even trivial customizations to malware code enabled it to bypass the sensors.  The industry began looking to machine learning for its ability to generate algorithms that generalize from known data in order to properly classify new and unknown data. The application of machine learning is not limited to malware detection.  As evidenced by Splunk’s acquisition of Caspida, a behavioral analytics company, the industry is seeing success in the use of algorithms to effectively classify and visualize the behavior of network elements. These developments give tier 1 security analysts the tools to perform at a higher level of skill.

With Great Potential Comes Great Challenges

Despite its great potential, getting machine learning to produce useful algorithms is no easy task.  One major challenge involves feature engineering.  Before the math can be let loose on a problem, data scientists need to determine the features of the problem that the model will analyze.  This is a naturally arduous process that requires an appropriate level of domain knowledge to be brought to task.  Domain knowledge in cyber security is varied and complex, requiring the data scientist to work closely with a network security expert.  In this case, feature design depends on the openness of communication between two types of individuals, both in short supply and each engaged in deep levels of thought within disparate technical disciplines.

Machine Learning Goes Deeper

Cyber security product vendors are starting to look toward a particular branch of machine learning that has been making impressive advances in recent years.  Deep learning is credited with giving computers the ability to correctly identify objects in photographic images, and the ability to parse meaning from natural human speech.

These deep learning models are based on “neural network” architectures, so called because they draw inspiration from models of the human brain. One of the processes replicated by deep learning is the automatic discovery of features significant to classifying data. In other words, deep learning methods remove the need for feature engineering.  This does not quite mean a free lunch since these methods have their own challenges and limitations. However, they can provide novel approaches to security problems that are better suited to a development team’s resources.  Already, companies are working with deep learning to identify unknown protocols or discover malware on enterprise networks.  Cyber security firm, Deep Instinct, prominently advertises its use of deep learning for endpoint protection.  At last year’s Black Hat conference, another endpoint protection provider, Cylance, demonstrated its research in converting code to bitmaps so that it could be analyzed by deep learning models.

The momentum in the field of deep learning should be a reassurance to innovators still looking to get involved. Growth of the field can be measured by one of its most prominent benchmark competitions, the ImageNet Large Scale Visual Recognition Challenge (ILSVRC).  The contest draws some of the biggest names in technology including IBM’s Watson Research Center, Google, Microsoft Research, and MIT, as well as many up and coming startups, competing to demonstrate deep-learning’s ability to surpass human capability in image and object recognition across several categories. Many of the category winners in last year’s competition were aided by deep-learning optimized platforms that featured NVIDIA GPUs to greatly reduce the time needed to develop and train deep learning neural networks. Three of the winning teams at ILSVRC 2015, including Microsoft Research and SenseTime, a leader in facial recognition for video surveillance applications, placed 1st in their respective categories supported by Deep Learning platforms developed by AMAX. By collaborating with AMAX, developers gained world class platforms both for use in their own ground breaking research and as security appliances to be deployed at customer sites in order to thwart the next generation of cyber threats. For more information on deep learning platforms or OEM services geared towards cyber security companies looking to bring an integrated security appliance to market, please contact AMAX to fast-track your development.

Rob Lundy is an independent technology and product marketing consultant with a decade of experience in hardware and software solutions for national defense and cyber security. He can be reached at

Posted in Cyber Security, Deep Learning, Server Appliance Manufacturing | Tagged , | Leave a comment