Neptune Monitoring System

[Muriel Trettin]

Tensof thousands of aquariums. Billions of measurements a year. Near zero downtime. Neptune Systems has been helping aquarists monitor their marine ecosystems for 25 years. In an aquarium, parameters can change in the blink of an eye. Apex provides 24/7 monitoring of your aquarium to help fine-tune the stability your fish and corals need to thrive while alerting you to any changes that might negatively impact their ability to survive.

For 25 years tens of thousands of aquarists have relied on Neptune for peace of mind. With Apex monitoring and control, Neptune users have rested easy knowing that if something goes wrong they will know about it and can act before it becomes a real problem.

Take peace of mind knowing the Apex is watching over your aquarium. The Apex will alert you via a mobile notification, email, or even sound alarm should parameters fall outside your desired range. Any aquarist knows that knowing is peace of mind.

Knowing about an issue before it becomes a problem is half the solution, but Neptune takes you further. With Apex control a broken heater can be automatically shut off or a drop in pH can trigger buffer dosing. The possibilities are endless.

When it comes to maintaining and monitoring your prized ecosystem you should accept nothing but the best. With over a decade of continuous online monitoring, Neptune network downtime has been measured in single minutes.

We are now transitioning all ordering of Neptune Systems, EcoTech Marine, AquaIllumination, Bulk Reef Supply Wholesale, and all of our Aperture Exclusive Distribution Brands to the new Aperture Business website (aperturebusiness.com).

Neptune has put together a package that included the salinity, temp, pH and ORP probes, along with WiFi connectivity all in a single package. The Apex Controller is by far one of the most advanced dedicated aquarium controllers on the market today. The wide variety of controllability with the Apex Controller System and its easy to use cloud interface makes it a perfect option for a fully automated reef tank. The Apex Controller has two different methods of control, through Fusion, the cloud-based software, and through the dashboard. The dashboard lets you always have access to your Apex in any type of event, power outage, or if the internet is down through a local area network (LAN), unlike many other cloud-based controllers. The Energy Bar 832 can also tell you how much power each individual outlet is consuming letting you set alarms, send emails, or even have IOS notifications let you know before things go wrong.

Included WiFi lets you now connect to the internet even easier. With the Fusion Cloud interface everything is easier from calibrating probes, to customizing outlets. Fusion gives you the ability to monitor and control your Apex from anywhere in the world.

The Energy Bars have been completely redesigned, with more robust outlets, individual LEDs show the status of each outlet, per outlet power consumption, and 1Link Accessories like the DOS and WAV pump can now connect directly to the Energy Bar 832 saving room, and eliminating messy cords keeping your tanks equipment looking perfect.

The new Apex's Head Unit now includes the ability to have salinity monitoring, pH, ORP, and Temperature monitoring right out of the box. The completely new design also incorporates a sleek, rotating mounting system letting you securely mount the brains of the Apex system while having easy access to probe cords, and aquabus connections.

Now with built-in WiFi, connecting your Apex system has never been easier. With the totally free Fusion cloud based software, getting connected, and controlling can be done within minutes. You can now do firmware updates through WiFi on your local network

The most comprehensive aquarium controlling and monitoring system to date is finally here. The Energy Bar 832 now has eight individual, 7 amp relayed outlets letting you use any outlet for any device. Each individual outlet can also monitor and detect power usage in amps or watts, letting you set alarms and send emails when pumps have clogged or failed, lights are not turning on, heaters leaking current, letting you be notified of the issue before it becomes a problem. With electricity costing more and more money per Kilowatt hour having the ability to monitor each pump, heater, and light lets you now make better decisions on the overall value of your equipment.

Each Energy Bar 832 also has two 24VDC accessory ports that can power small pumps, solenoids, fans and more, while the built-in 100W DC power supply lets you control up to 3 1Link accessories like the WAV circulation pumps, and DOS fluid metering system without the need for any additional power supplies or consuming additional outlets. The EB832 also has 3 AquaHub connections letting you expand to any existing or upcoming module making your Apex System truly limitless.

Neptune didn't stop there, with an all new mounting system on the Energy Bar orients the power strip in the correct direction keeping cords neat and tidy. While the outlets have a bit wider spacing, and orientated opposite of each other keeping room for those bulky transformers, allowing you to plug in almost anything you want, where you want.

Double junction pH, and ORP probes lets your readings stay more stable without the need to calibrate as often. With the Apex System calibrating your probes through your dashboard is easier than ever before. The dashboard will automatically recognize when your readings have stabilized and letting you know when to move to the next step, no more guesswork and reading RAW data.

Fusion is a cloud-based site that lets you access your Apex from anywhere in the world. Letting you customize outlets, set reminders, log testing parameters, while monitoring and controlling any aspect of your aquarium through your phone, tablet, or desktop.

Apex Fusion has released an IOS app that now integrates directly with your Apex Controller System. Having the app now lets your Apple desktop, Ipad, or phone receive notifications directly from your Apex Controller. No more need for emails or text messages.

In terms of monitoring your application, you might be worried about system metrics, error rates, traffic volume, app loading times, infrastructure (server counts, load, CPU/GPU usage), things like that. Your application scales to a lot of users, and it works as intended and solves the problem it was built to solve.

From the moment you deploy your model to production, it begins to degrade in terms of performance. A model is at its best just before being deployed to production. This is why deployment should not be your final step.

There could also be situations where you deploy models that are localized to multiple geographic regions. How do you ensure that your models serve the intended results (predictions) to each client concerning high interest in their own reality? How do you explain why a model makes certain predictions for one geographic region and a different set of predictions for another region?

This should give you a broad idea of the challenges that you may encounter after deploying to prod, and why you need to continue your good work after deployment by monitoring your model(s) in production.

The primary goal of monitoring here is to flag any data quality issue, either from the client or due to an unhealthy data pipeline, before the data is sent to your model (which would generate unreliable predictions in response).

For example, a schema change could be the database admin for one of your data sources renaming a feature column and adding another column to catch new data. While this might scale through the preprocessing step, it sure would be a hassle for the model to predict, so it will likely give a partial response as the prediction. The schema change would mean that the model needs to be updated before it can map the relationship between the new feature column and the old ones.

Feature drift can occur due to data quality issues or general changes in the real world, like changes in the preferences of business customers. An example of feature/attribute drift is illustrated below, where a historical set of attributes are used as a baseline and newer attributes are compared so that changes in the distribution of the attributes can be detected.

You can monitor model versions by configuring your retraining pipeline to automatically report model versions after training and log metadata to a metadata store. Version history should be logged to an evaluation store alongside model predictions, this way problems will be easier to tie to model versions.

Every business faces security threats. With machine learning applications increasingly becoming the central decision system of most companies, you have to be concerned about the security of your model in production. Most machine learning models are susceptible to adversarial attacks.

Concerted adversaries come directly from systems or hackers that intentionally engage your system through adversarial attacks. They mislead the system with adversarial examples so it can provide unreliable results and cause it to make mistakes. Such attacks (which are often anomalous) represent a specific problem in the safety of your machine learning application in production and need to be monitored.

In business-critical applications, speed matters. How fast experts detect an adversarial threat, study such a threat, patch the vulnerability by retraining the model, and redeploy the model may make all the difference for the business.

Monitoring model output in production is not just the best indicator of model performance, but it also tells us if business KPIs are being met. In terms of model predictions, the most important thing to monitor is model performance in line with business metrics.

Using metrics to evaluate model performance is a big part of monitoring your model in production. Different metrics can be used here, such as classification, regression, clustering, reinforcement learning, and so on.

3a8082e126