Tesla .ssq file?

Tonight, I was a large download by the car, and saw that it was a .ssq file. The file name is consistent with the firmware naming convention, but I am not sure on what it is. The file itself is 5.11 GB, and in my case its name starts with “NA”. I am guessing, this might be the maps its updating.

Below are a couple of screenshots showing this. I am trying to make sense of the binary file, but not making much headway.

Curious, anyone has any ideas?

Update: I found out what .ssq files are; read up more here.

Neural Network – Cheat Sheet

Neural Networks, today, help in a great set of tasks, that until very recently wasn’t possible at all – be it from computer vision, to medical diagnosis, to speech translation and forms a key cornerstone to a lot of ‘magic’ that Machine Learning and AI offers today.

I did blog about Neural Network types (and MarI/O) sometime back; I surely cannot take credit for creating these three cheat sheets but they are awesome and hope you get to use and enjoy them too.

Neural Network Graphs

Clearing out Windows 10 command prompt history

My command prompt history is quite long, and a lot over time is not essentially garbage. I was looking at a way to clean it out. Most of the solutions online I found were not correct – I don’t know if things changed over time, but the latest version of Windows I am on (Windows 10 Pro 1803), it did not work.

So, here are two ways that you can do this. One is using the registry editor (RegEdit), and the other is running a simple script that you can either copy and paste from below or you can download and run it.

If you are going to be using RegEdit, and living dangerously then Press WinKey + R and type “regedit” (without quotes) and press enter to get the Registry Editor going as shown below.

Run command to start Registry Editor

And on the new Windows navigate to the following key: HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\RunMRU and delete that. You can right click on the key name and choose delete.

It is important to double check because if you miss it, or delete something else, there is no recovery. (Why do you think I was saying, you like to live dangerously). See the screenshot below.

NOTE: It is always recommended to backup the registry before doing this, so at least you could restore it back to the state. To backup select File -> Export.

A better way, and less dangerous would be to run the following script in a elevated command prompt (i.e. a Admin command prompt) which will do the same thing, but more safer. You can just copy the command from below and paste it. Or alternatively you can download this simple script and run it locally (also from a elevated command prompt).

reg delete "HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\RunMRU" /f

Tesla debug/diagnostic screens

I don’t know how to get to debug / dev mode on a Tesla, but did come across this old post, on how someone was in a test drive, which did  have this mode.

Now this is quite old, so a lot has changed, but am impressed that a lot of the components and foundational architecture was setup. I am particularly impressed how each cell in the battery pack and report its state. The BMS that you see is the Battery Management System – that firmware is separate from the car’s firmware.

Tesla diagnostic screen

You can see more photos and geek out online here.

And of course if you really want to geek out, then check out su-tesla, where Hemera has really has gone to party. I don’t know how to do this, and I have a lot of respect for Hemera to do this – she has a lot of guts. Also not sure what the wife would think about it and kick me out. Maybe. 🙂

I am curious though, if those ‘custom’ Ethernet connectors are M12 connectors (PDF) which are quite standard in some industries. Even Amazon sells cables for them.

And finally, from a more (relatively) recent update, the AutoPilot has a tremendous amount of data. As reported here, and you can see on the video below, the volume of data is massive, and quite interesting. For example, what decides there are 4 virtual lanes? The car below is a US car (the country code 840 is a ISO 3166 code).

Thought of the day

Beware of programmers that carry screwdrivers

– Unknown

Tesla voice command list

As I was trying to understand more on the capabilities of the car, and what options I can do. The voice recognition on the car is quite impressive, it does seem as good at understanding as Amazon’s Echo, at least in the early days of “Alexa” (but that is a different story for another time).

I was trying to understand what things can I control, or the options one has via the voice. I am still not used to it, and keep forgetting, that is a option especially when driving. As of the v8 firmware series, the following are the choices that work for voice. Credit to Ingineer for discovering the full list when hacking the car.

The options in English are listed below, and this is missing the “ho ho ho” Easter egg and also the “cancel navigation” command.

"voice_command_list" : [
"command_type" : "navigate",
"description" : "drive to",
"description" : "drive 2",
"description" : "dr to",
"description" : "dr 2",
"description" : "drive",
"description" : "dr",
"description" : "navigate to",
"description" : "navigate 2",
"description" : "navigate",
"description" : "where is",
"description" : "take me to",
"description" : "take me 2",
"description" : "take me",

"command_type" : "call",
"description" : "call",
"description" : "dial",
"description" : "phone",

"command_type" : "note",
"description" : "note",
"description" : "report",
"description" : "bug note",
"description" : "bug report",

"command_type" : "play",
"description" : "play",
"description" : "plays",
"description" : "listen to",
"description" : "listens to",
"description" : "listen 2",
"description" : "listens 2"
]

And if you are keen to know, these are stored as a json file internally, and the fill list here here:

{
    "voice_command_list" : [
        {
            "command_type" : "navigate",
            "description" : "drive to",
            "command_regexp" : "^drive to\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "drive 2",
            "command_regexp" : "^drive 2\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "dr to",
            "command_regexp" : "^dr to\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "dr 2",
            "command_regexp" : "^dr 2\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "drive",
            "command_regexp" : "^drive\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "dr",
            "command_regexp" : "^dr\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "navigate to",
            "command_regexp" : "^navigate to\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "navigate 2",
            "command_regexp" : "^navigate 2\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "navigate",
            "command_regexp" : "^navigate\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "where is",
            "command_regexp" : "^where is\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "take me to",
            "command_regexp" : "^take me to\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "take me 2",
            "command_regexp" : "^take me 2\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "take me",
            "command_regexp" : "^take me\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "naviguer à",
            "command_regexp" : "^naviguer à\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "naviguer au",
            "command_regexp" : "^naviguer au\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "aller à",
            "command_regexp" : "^aller à\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "aller au",
            "command_regexp" : "^aller au\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "nach navigieren",
            "command_regexp" : "^nach\\b(.*)\\bnavigieren$"
        },
        {
            "command_type" : "navigate",
            "description" : "zur navigieren",
            "command_regexp" : "^zur\\b(.*)\\bnavigieren$"
        },
        {
            "command_type" : "navigate",
            "description" : "zu navigieren",
            "command_regexp" : "^zu\\b(.*)\\bnavigieren$"
        },
        {
            "command_type" : "navigate",
            "description" : "nach fahren",
            "command_regexp" : "^nach\\b(.*)\\bfahren$"
        },
        {
            "command_type" : "navigate",
            "description" : "zur fahren",
            "command_regexp" : "^zur\\b(.*)\\bfahren$"
        },
        {
            "command_type" : "navigate",
            "description" : "zu fahren",
            "command_regexp" : "^zu\\b(.*)\\bfahren$"
        },
        {
            "command_type" : "navigate",
            "description" : "wo ist",
            "command_regexp" : "^wo ist\\b(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "navigiere nach",
            "command_regexp" : "^navigiere nach\\b(.*)\\b$"
        },
        {
            "command_type" : "navigate",
            "description" : "navigiere zu",
            "command_regexp" : "^navigiere zu\\b(.*)\\b$"
        },
        {
            "command_type" : "navigate",
            "description" : "导航到",
            "command_regexp" : "^导航到(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "在哪",
            "command_regexp" : "^(.*)在哪$"
        },
        {
            "command_type" : "navigate",
            "description" : "开车到",
            "command_regexp" : "^开车到(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "导航去",
            "command_regexp" : "^导航去(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "導航去",
            "command_regexp" : "^導航去(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "導航到",
            "command_regexp" : "^導航到(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "帶我去",
            "command_regexp" : "^帶我去(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "帶我到",
            "command_regexp" : "^帶我到(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "去",
            "command_regexp" : "^去(.*)$"
        },
        {
            "command_type" : "navigate",
            "description" : "到",
            "command_regexp" : "^到(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "call",
            "command_regexp" : "^call\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "dial",
            "command_regexp" : "^dial\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "phone",
            "command_regexp" : "^phone\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "appeler",
            "command_regexp" : "^appeler\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "composer",
            "command_regexp" : "^composer\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "wählen",
            "command_regexp" : "^(.*)\\bwählen$"
        },
        {
            "command_type" : "call",
            "description" : "anrufen",
            "command_regexp" : "^(.*)\\banrufen$"
        },
        {
            "command_type" : "call",
            "description" : "wähle",
            "command_regexp" : "^wählen\\b(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "ruf an",
            "command_regexp" : "^ruf\\b(.*)\\ban$"
        },
        {
            "command_type" : "call",
            "description" : "rufe an",
            "command_regexp" : "^rufe\\b(.*)\\ban$"
        },
        {
            "command_type" : "call",
            "description" : "打电话给",
            "command_regexp" : "^打电话给(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打电话给",
            "command_regexp" : "^给(.*)打电话$"
        },
        {
            "command_type" : "call",
            "description" : "拨打",
            "command_regexp" : "^拨打(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打给",
            "command_regexp" : "^打给(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打電話俾",
            "command_regexp" : "^打電話俾(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打俾",
            "command_regexp" : "^打俾(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打電話去",
            "command_regexp" : "^打電話去(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打去",
            "command_regexp" : "^打去(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打電話比",
            "command_regexp" : "^打電話比(.*)$"
        },
        {
            "command_type" : "call",
            "description" : "打比",
            "command_regexp" : "^打比(.*)$"
        },
        {
            "command_type" : "note",
            "description" : "note",
            "command_regexp" : "^note\\b(.*)$"
        },
        {
            "command_type" : "note",
            "description" : "report",
            "command_regexp" : "^report\\b(.*)$"
        },
        {
            "command_type" : "note",
            "description" : "bug note",
            "command_regexp" : "^bug note\\b(.*)$"
        },
        {
            "command_type" : "note",
            "description" : "bug report",
            "command_regexp" : "^bug report\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "play",
            "command_regexp" : "^play\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "plays",
            "command_regexp" : "^plays\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "listen to",
            "command_regexp" : "^listen to\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "listens to",
            "command_regexp" : "^listens to\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "listen 2",
            "command_regexp" : "^listen 2\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "listens 2",
            "command_regexp" : "^listens 2\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "écouter",
            "command_regexp" : "^écouter\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "jouer",
            "command_regexp" : "^jouer\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "spielen",
            "command_regexp" : "^(.*)\\bspielen$"
        },
        {
            "command_type" : "play",
            "description" : "hören",
            "command_regexp" : "^(.*)\\bhören$"
        },
        {
            "command_type" : "play",
            "description" : "abspielen",
            "command_regexp" : "^(.*)\\babspielen$"
        },
        {
            "command_type" : "play",
            "description" : "abhören",
            "command_regexp" : "^(.*)\\babhören$"
        },
        {
            "command_type" : "play",
            "description" : "spiele",
            "command_regexp" : "^spiele\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "spiel",
            "command_regexp" : "^spiel\\b(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "播放",
            "command_regexp" : "^播放(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "收听",
            "command_regexp" : "^收听(.*)$"
        },
        {
            "command_type" : "play",
            "description" : "我想聽",
            "command_regexp" : "^我想聽(.*)$"
        }
    ]
}

Whilst the #NLP engine working on this is quite good, and impressive, I am hopeful that there will be more options added. Elon did share that is something they are working on, and it might be part of the updated v9 release coming out in the next few weeks.

How many lines of code does it take?

Often once hears are Lines of Code (LoC) as a metric. And for you to get a sense of what it means, below is a info-graphic that outlines some popular products, and services and the LoC that takes. Always interesting to get perspective – either appreciate some home grown system you are managing, or worried about a stinking pile you are going to inherit or build. 🙂

Generating Tesla authentication token – cURL script

I did write a simple Windows (desktop) app called TeslaTokenGenerator, for those who wanted to create authentication tokens for their Tesla, and use with 3rd party apps/data loggers. 

TeslaTokenGenerator can also create a cURL script for you to use, if you prefer not wanting to install anything. It is easy to find this online, but some of you have pinged me to get more details on this. So, I have the script below that you can use. Once you copy this, you will need to update your Tesla account login details (email and password) and run it in a console (command line) and it will all the same API’s to create the token, which then you can save.

curl -X POST -H "Cache-Control: no-cache" -H "Content-Type: multipart/form-data; boundary=----WebKitFormBoundary7MA4YWxkTrZu0gW" -F "grant_type=password" -F "client_id=81527cff06843c8634fdc09e8ac0abefb46ac849f38fe1e431c2ef2106796384" -F "client_secret=c7257eb71a564034f9419ee651c7d0e5f7aa6bfbd18bafb5c5c033b093bb2fa3" -F "email=YOUR-TESLA-LOGIN-EMAIL@SOMEWHERE.COM" -F "password=YOUR-TESLA-ACCOUNT-PASSWORD" "https://owner-api.teslamotors.com/oauth/token"

You can see the screenshots of this below too – one in Windows, and another in Linux (well Bash on Windows, but it is real Linux).

My Tesla Model 3 “Keyfob”

Inspired by a few folks on a few forums online, I took the liberty to extend their idea using a IoT Button, that acts as a simple “keyfob” for the Model 3.

The main goal was being to allow my daughter to lock and unlock the car at home. She is too young to have a phone, and without a more traditional fob, this gets a little annoying. 

I extended the original idea, to understand the different presses (Single, Double, and Long press), and accordingly called the appropriate API to lock the car (single press – think of it as a single click), unlock on a double press, and open the charge port on a long press (when one presses and holds the button 2-3 secs).

Model 3 “Keyfob”

For those who aren’t aware, the Amazon IoT buttons calls a Lambda function on AWS and plugging into that, one can extend this. The button needs to be connected, and online for this to work, and in my case, it is on the home wifi network. 

Update: Many of you asked on how to set this up for yourself; I got around to blogging all the step on that; you can read those here.

Windows Tesla Auth Token Generator

If you have a Tesla, and are using (or wanting to use) 3rd party tools or data loggers, the one think they of course need is to authenticate your details with Tesla. A simple, but insecure way is to use your Tesla credentials – and surprisingly many people just happily share and use this.

I wasn’t comfortable doing this – after-all, they have access to your account where you can control a lot of things. Also, there are a few online tools that can generate the auth token, but again I wasn’t comfortable, as I did not know what they saved, or what they did not. 🙂

So, I wrote a simple Windows app that can allow you to generate a auth token that you can save. The application itself is simple. You enter your Tesla credentials, click on Generate Token and can save the generated token. 

Tesla Token Generator Application

To test, if the generated token is working – click on the Test Token button. If everything is working as expected, you will see a list of vehicles that is associated with your account.

If you prefer to use the cURL script, click on the Generate cURL, will generate this and copy it to your clipboard. And it works across operating systems as you can see below (Windows, and Linux), but should also work on Mac.

I do intent to open source this, so folks can have a look at the code, and the Tesla REST APIs. Until then you can download the setup from here.

Leave a comment if you have any issues or any requests.

Update: v1.0.1 Published with minor updates. You can download from the same link above. This adds the revoke screen and some house keeping.

The merits of #AI

Thought of the week:

Artificial Intelligence stands no chance against natural Stupidity.

#ArtificalIntelligence

#ML training data

Seem like my training data for the car – perhaps a hint of #bias. 😂

#GeekyJokes #ML #AIJokes

Neural network basics–Activation functions

Neural networks have a very interesting aspect – they can be viewed as a simple mathematical model that define a function. For a given function f(x) which can take any input value of x, there will be some kind a neural network satisfying that function. This hypothesis was proven almost 20 years ago (“Approximation by Superpositions of a Sigmoidal Function” and “Multilayer feedforward networks are universal approximators”) and forms the basis of much of #AI and #ML use cases possible.

It is this aspect of neural networks that allow us to map any process and generate a corresponding function. Unlike a function in Computer Science, this function isn’t deterministic; instead is confidence score of an approximation (i.e. a probability). The more layers in a neural network, the better this approximation will be.

In a neural network, typically there is one input layer, one output layer, and one or more layers in the middle. To the external system, only the input layer (values of x), and the final output (output of the function f(x)) are visible, and the layers in the middle are not and essentially hidden.

Each layer contains nodes, which is modeled after how the neurons in the brain works. The output of each node gets propagated along to the next layer. This output is the defining character of the node, and activates the node to pass on its value to the next node; this is very similar to how a neuron in the brain fires and works passing on the signal to the next neuron.

Neural Network
Neural Network

To make this generalization of function f(x) outlined above to hold, the that function needs to be continuous function. A continuous function is one where small changes to the input value x, creates small changes to the output of f(x). If these outputs, are not small and the value jumps a lot then it is not continuous and it is difficult for the function to achieve the approximation required for them to be used in a neural network.

For a neural network to ‘learn’ – the network essentially has to use different weights and biases that has a corresponding change to the output, and possibly closer to the result we desire. Ideally small changes to these weights and biases correspond to small changes in the output of the function. But one isn’t sure, until we train and test the result, to see that small changes don’t have bigger shifts that drastically move away from the desired result. It isn’t uncommon to see that one aspect of the result has improved, but others have not and overall skewing the results.

In simple terms, an activation function is a node that attached to the output of a neural network, and maps the resulting value between 0 and 1. It is also used to connect two neural networks together.

An activation function can be linear, or non-linear. A linear isn’t terribly effective as its range is infinity. A non-linear with a finite range is more useful as it can be mapped as a curve; and then changes on this curve can be used to calculate the difference on the curve between two points.

There are many times of activation function, each either their strengths. In this post, we discuss the following six:

  • Sigmoid
  • Tanh
  • ReLU
  • Leaky ReLU
  • ELU
  • Maxout

1. Sigmoid function

A sigmoid function can map any of input values into a probability – i.e., a value between 0 and 1. A sigmoid function is typically shown using a sigma (\sigma). Some also call the (\sigma) a logistic function. For any given input value,  x the official definition of the sigmoid function is as follows:

\sigma(x) \equiv \frac{1}{1+e^{-x}}

If our inputs are x_1, x_2,\ldots, and their corresponding weights are w_1, w_2,\ldots, and a bias b, then the previous sigmoid definition is updated as follows:

\frac{1}{1+\exp(-\sum_j w_j x_j-b)}

When plotted, the sigmoid function, will look plotted looks like this curve below. When we use this, in a neural network, we essentially end up with a smoothed out function, unlike a binary function (also called a step function) – that is either 0, or 1.

For a given function, f(x), as x \rightarrow \infty, f(x) tends towards 1. And, as as x \rightarrow -\infty, f(x) tends towards 0.

Sigmoid function
Sigmoid function

And this smoothness of \sigma is what will create the small changes in the output that we desire – where small changes to the weights (\Delta w_j), and small changes to the bias (\Delta b) will produce a small changes to the output (\Delta output).

Fundamentally, changing these weights and biases, is what can give us either a step function, or small changes. We can show this as follows:

\Delta \mbox{output} \approx \sum_j \frac{\partial \, \mbox{output}}{\partial w_j} \Delta w_j + \frac{\partial \, \mbox{output}}{\partial b} \Delta b

 

One thing to be aware of is that the sigmoid function suffers from the vanishing gradient problem – the convergence between the various layers is very slow after a certain point – the neurons in previous layers don’t learn fast enough and are much slower than the neurons in later layers. Because of this, generally a sigmoid is avoided.

2. Tanh (hyperbolic tangent function)

Tanh, is a variant of the sigmoid function, but still quite similar – it is a rescaled version and ranges from –1 to 1, instead of 0 and 1. As a result, its optimization is easier and is preferred over the sigmoid function. The formula for tanh, is

\tanh(x) \equiv \frac{e^x-e^{-z}}{e^X+e^{-x}}

Using, this we can show that:

\sigma(x) = \frac{1 + \tanh(x/2)}{2}.

Sigmoid vs Tanh
Sigmoid vs Tanh

Tanh also suffers from the vanishing gradient problem. Both Tanh, and, Sigmoid are used in FNN (Feedforward neural network) – i.e. the information always moves forward and there isn’t any backprop.

FNN
FNN

3. Rectified Linear Unit (ReLU)

A rectified linear unity (ReLU) is the most popular activation function that is used these days.

\sigma(x) = \begin{cases} x & x > 0\\ 0 & x \leq 0 \end{cases}

 

ReLU’s are quite popular for a couple of reasons – one, from a computational perspective, these are more efficient and simpler to execute – there isn’t any exponential operations to perform. And two, these doesn’t suffer from the vanishing gradient problem.

ReLU
ReLU

The one limitation ReLU’s have, is that their output isn’t in the probability space (i.e. can be >1), and can’t be used in the output layer.

As a result, when we use ReLU’s, we have to use a softmax function in the output layer.  The output of a softmax function sums up to 1; and we can map the output as a probability distribution.

\sum_j a^L_j = \frac{\sum_j e^{z^L_j}}{\sum_k e^{z^L_k}} = 1.

 

Another issue that can affect ReLU’s is something called a dead neuron problem (also called a dying ReLU). This can happen, when in the training dataset, some features have a negative value. When the ReLU is applied, those negative values become zero (as per definition). If this happens at a large enough scale, the gradient will always be zero – and that node is never adjusted again (its bias. and, weights never get changed) – essentially making it dead! The solution? Use a variation of the ReLU called a Leaky ReLU.

4. Leaky ReLU

A Leaky ReLU will usually allow a small slope \alpha on the negative side; i.e that the value isn’t changed to zero, but rather something like 0.01. You can probably see the ‘leak’ in the image below. This ‘leak’ helps increase the range and we never get into the dying ReLU issue.

ReLU vs. Leaky ReLU
ReLU vs. Leaky ReLU

5. Exponential Linear Unit (ELU)

Sometimes a ReLU isn’t fast enough – over time, a ReLU’s mean output isn’t zero and this positive mean can add a bias for the next layer in the neural network; all this bias adds up and can slow the learning.

Exponential Linear Unit (ELU) can address this, by using an exponential function, which ensure that the mean activation is closer to zero. What this means, is that for a positive value, an ELU acts more like a ReLU and for negative value it is bounded to -1 for \alpha = 1 – which puts the mean activation closer to zero.

\sigma(x) = \begin{cases} x & x \geqslant 0\\ \alpha (e^x - 1) & x < 0\end{cases}

 

When learning, this derivation of the slope is what is fed back (backprop) – so for this to be efficient, both the function and its derivative need to have a lower computation cost.

elu-and-relu
ELU vs ReLU

And finally, there is another various of that combines with ReLU and a Leaky ReLU called a Maxout function.

So, how do I pick one?

Choosing the ‘right’ activation function would of course depend on the data and problem at hand. My suggestion is to default to a ReLU as a starting step and remember ReLU’s are applied to hidden layers only. Use a simple dataset and see how that performs. If you see dead neurons, than use a leaky ReLU or Maxout instead. It won’t make sense to use Sigmoid or Tanh these days for deep learning models, but are useful for classifiers.

In summary, activation functions are a key aspect that fundamentally influence a neural network’s behavior and output. Having an appreciation and understanding on some of the functions, is key to any successful ML implementation.

Netron – deep learning and machine learning model visualizer

I was looking at something else and happen to stumble across something called Netron, which is a model visualizer for #ML and #DeepLearning models. It is certainly much nicer than for anything else I have seen. The main thing that stood out for me, was that it supports ONNX , and a whole bunch of other formats (Keras, CoreML), TensorFlow (including Lite and JS), Caffe, Caffe2, and MXNet. How awesome is that?

This is essentially a cross platform PWA (progressive web app), essentially using Electron (JavaScript, HTML5, CSS) – which means it can run on most platforms and run-times from just a browser, Linux, Windows, etc. To debug it, best to use Visual Studio Code, along with the Chrome debugger extension.

Below is a couple of examples, of visualizing a ResNet-50 model – you can see both the start and the end of the visualization shown in the two images below to get a feel of things.

image
Start of ResNet-50 Model

image
End of ResNet-5o model

And some of the complex model seem very interesting. Here is an example of a TensorFlow Inception (v3) model.

image

And of course, this can get very complex (below is the same model, just zoomed out more).

image

I do think it is a brilliant, tool to help understand the flow of things, and what can one do to optimize, or fix. Also very helpful for folks who are just starting to learn and appreciate the nuances.

The code is released under a MIT license and you can download it here.

Machine learning use-cases

Someone recently asked me, what are some of the use cases / examples of machine learning. Whilst, this might seem as an obvious aspect to some of us, it isn’t the case for many businesses and enterprises – despite that they uses elements of #ML (and #AI) in their daily life – as a consumer.

Whilst, the discussion gets more interesting based on the specific domain and the possibly use cases (of course understanding that some might not be sure f the use case – hence the question in the first place). But, this did get me thinking and wanted to share one of the images we use internally as part of our training that outcomes some of the use cases.

Machine Learning Use Cases
Machine Learning Use Cases

These are not 1:1 and many of them can be combined together to address various use cases – for example a #IoT device sending in a sensor data, that triggers a boundary condition (via a #RulesEngine), that in addition to executing one or more business rule, can trigger a alert to a human-in-the-loop (#AugmentingWorkforce) via a #DigitalAssistant (say #Cortana) to make her/him aware, or confirm some corrective action and the likes. The possibilities are endless – but each of these elements triggered by AI/ML and still narrow cases and need to be thought of in the holistic picture.

Synthetic Sound

Trained a model to create a synthetic sound that sounds like me. This is after training it with about 30 sentences – which isn’t a lot.

To create a synthetic voice, you enters some text, using which is then “transcribed” using #AI and your synthetic voice is generated. In my case, at first I had said AI, which was generated also as “aeey” (you can have a listen here). So for the next one, changed the AI to Artificial Intelligence.

One does need to be mindful of #DigitalEthics, as this technology improves further. This is with only a very small sampling of data. Imagine what could happen, with public figures – where their recordings are available quite easily in the public domain. I am thinking the ‘digital twang’ is one of the signatures and ways to stamp this as a generated sound.

My self-driving car

Over the last few weeks, I built a self-driving car – which essentially is a remote control Rx car that uses a raspberry pi running Python, TensorFlow implementing a end-to-end convolution neural network (CNN)

Of course other than being  a bit geeky, I do think this is very cool to help understand and get into some of the basic constructs and mechanics around a number of things – web page design, hardware (maker things), and Artificial Intelligence principles.

There are two different models here – they do use the same ASC and controller that can be programmed. My 3D printer, did mess up a little (my supports were a little off) and which is why you see the top not clean.

The sensor and camera are quite basic, and there is provisions to add and do better over time. The Pi isn’t powerful enough to train the model – you need another machine for that (preferably a I7 core with a GPU). Once trained you can run the model on the Pi for inference.

This is the second car, which is a little different hardware, but the ESC to control the motor and actuators are the same.

The code is simple enough; below is an example of the camera (attached) to the Pi, saving the images it is seeing. Tubs is the location where the images are saved; these can then be transferred to another machine for training or inference.

import donkey as dk

#initialize the vehicle
V = dk.Vehicle()

#add a camera part
cam = dk.parts.PiCamera()
V.add(cam, outputs=['image'], threaded=True)

#add tub part to record images
tub = dk.parts.Tub(path='~/d2/gettings_started', 
                   inputs=['image'], 
                   types=['image_array'])
V.add(tub, inputs=inputs)

#start the vehicle's drive loop
V.start(max_loop_count=100)

Below you can see the car driving itself around the track, where it had to be trained first. The reason it is not driving perfectly is because during training (when I was manually driving it around), I crashed a few times and as a result the training data was messed up. Needed more time to clean that up and retrain it.

This is based on donkey car – which is an open source DIY for platform for small-scale self driving cars. I think it is also perfect to get into with those who have teenagers and a little older kids to get in and experiment. You can read up more details on how to go about building this, and the parts needed here.