Showing posts with label Neural Networks. Show all posts
Showing posts with label Neural Networks. Show all posts

Friday, 28 August 2026

Killing the LSTM: Why DANI is Pivoting to a Spiking Neural Network

After some extensive research and a few hard encounters with hardware reality, I’ve made a major architectural decision regarding DANI's brain: I am completely ripping out the LSTM neural network and replacing it with a Recurrent Spiking Neural Network (RSNN).

Given my recent enthusiasm for the LSTM architecture, this will probably come as a surprise to anyone following along. However, I have my reasons, and as usual, they come down to math, metal, and thermal throttling.

What is wrong with LSTMs and the Traditional ANN?

Traditional Artificial Neural Networks (ANNs) are fantastic at what they were designed for: probability scoring. They allow us to simulate decision-making using massive matrix calculations mixed with a healthy dose of calculus.

But beneath the hood, they are brute-force mechanisms.

When a traditional ANN is fed data, that data cascades through every single neuron in the network. Even when a neuron has a zero value, the CPU still dutifully performs the mathematical operation of multiplying by zero. If a network has a shape of 10 inputs, 10 outputs, and 5 layers of 12 neurons each, that is 816 floating-point calculations just for one single feed-forward pass. If you start adding in simple recurrence, we are immediately up to 1,488 calculations.

Now, scale this up to something that would actually be useful for DANI. We need at least 100 inputs and about 64 outputs. With hidden layers of around 125 neurons each, we jump to 150,375 calculations per pass. To achieve an LSTM architecture, we can estimate the computation to be roughly four times that: over 600,000 calculations per tick.

As you can see, this grows exponentially. More importantly, it keeps the CPU running permanently hot. Don’t forget, DANI is powered by a Raspberry Pi 5. There is no GPU offloading here. I did manage to get DANI’s LSTM operating at 10Hz with 3.5 million parameters, but it required the Pi 5 to do absolutely nothing else, running flat-out at 100% utilization. DANI's brain was essentially doubling as a space heater, even when he was just "dreaming."

Once you add my simulated hormone system into the mix, the compute overhead goes up even further. I was basically following the same brute-force model that the big AI companies use, just without the luxury of a multi-million dollar server farm.

And then there is the training. To train an ANN of this style requires layer-by-layer calculus (backpropagation). This is heavy, blocking work. It takes a considerable amount of time, meaning we simply cannot train the network in real-time on anything substantial.

The Elegance of the Recurrent Spiking Neural Network

The architecture of an RSNN is topographically similar to other neural networks, but the execution is fundamentally different. It mimics biological reality much closer.

Instead of passing continuous floating-point numbers, each neuron has a membrane potential. As it receives signals from upstream neurons, that potential increases (or decreases, if the synapse is inhibitory). The neuron does absolutely nothing until that potential reaches a specific threshold. Once it hits the limit, it fires—or "spikes." And it spikes at a binary full power. The neuron then resets, either dropping to zero or subtracting the threshold from its current potential.

If a neuron doesn’t reach the threshold, it doesn’t spike. Period.

This means we only need to calculate the pathways for neurons that actually receive a spike. The computational reduction is staggering. The general consensus in the field is that for any given signal, only 5% to 10% of a spiking network is active, compared to the 100% density of a matrix-style network.

Training also becomes radically simpler. We only need to reinforce or weaken the synapses of the pathways that actually fired (a process akin to Spike-Timing-Dependent Plasticity). We apply a localized effector to those specific connections—no massive, network-wide calculus required.

But here is the real kicker for a systems engineer: an RSNN allows us to completely ditch floating-point math. By using integer mathematics and clever bit-shift operations (which are essentially free in terms of CPU cycles), we can completely bypass the messy, cycle-heavy floating-point multiplications.

How does this help DANI?

Obviously, DANI will not have to run so hot. If he has a quiet mind, the network will physically go quiet. When DANI rests, he will literally be saving power. By utilizing temporal calculations, we can also train him in real-time with near-zero impact on the system.

Furthermore, DANI’s brain becomes an event-driven architecture rather than a strict polling loop. Instead of processing rigidly on a timer, he will think and react in time with the world around him.

I will still be using the hormonal system I designed, but the key difference is how it integrates. The effective values of the hormones will now directly modulate the firing thresholds of the neurons. This allows his "moods" to physically alter the state of his brain and affect his behavior, even if his underlying memories remain unchanged.

I suppose I had better be nice to him.

Rethinking Asimov’s Laws


This pivot finally allows me to revisit a question I was pondering a while back: How do we encode Asimov’s Laws into an AI?

The answer is: we don’t. It’s as simple as that. We don’t hardcode the laws; we teach them.

By running an RSNN that learns in real-time through stimulus and reinforcement, the laws can become an emergent part of his personality. He will understand that hurting people is bad and that following human orders brings a reward, not because a line of code forces him to, but because his synapses have shaped themselves around those experiences.

So, I guess I really am going to have to be a parent to a robot.

Tuesday, 18 August 2026

Major Milestone: Skink Has Officially Bootstrapped!

Woohoo! It’s time for a long-overdue update on the language powering the future of DANI.

We just crossed a monumental threshold with Skink: it is now officially bootstrapped!

For anyone unfamiliar with compiler development, bootstrapping means that the Skink compiler can now compile its own source code. From here on out, all future versions of the Skink compiler will be written in Skink itself. It’s a huge rite of passage for any programming language project, and seeing it actually work after so much foundation work is insanely satisfying.

Where Go Fits In (For Now)

The handover
Now, does this mean I’m throwing Go out the window today? Not quite!

Go has a proven, rock-solid track record, and I’ll be keeping it around for prototyping work. Whenever I want to test out an idea quickly and I’m not sure if Skink is quite ready to handle the heavy lifting yet, Go gives me a reliable safety net.

However, the long-term vision hasn't changed: eventually, I plan to fully commit to Skink for almost all development across my projects, experiments, and research.

The Road to Skink 1.0

While bootstrapping is a massive step forward, we aren't quite at version 1.0 just yet. To earn that release tag, Skink needs to hit a few crucial targets:

  1. Cross-Platform Support: Clean cross-compilation across Windows, macOS, and Linux.
  2. Architecture Support: Full targeting for both x86 and ARM architectures.
  3. Embedded Target Transpilation: The ability to compile (or at least transpile down) to low-power platforms like the K210, ESP32, and standard Arduino C/C++.
  4. Direct Hardware I/O: First-class hardware abstraction libraries for lower-level protocols—specifically I2C, SPI, GPIO, and related interfaces.

Acceleration, Compute, and Graphics Backends

Beyond microcontrollers and standard CPUs, getting high-performance compute and neural inference support hooked up is high on the priority list.

  • CUDA Support: I still need to get CUDA fully enabled and tested. Because of my current local setup, this will likely require either acquiring new dedicated NVIDIA hardware or setting up a cloud-based VM for build and test pipelines.

  • Vulkan & AMD: My main local system currently relies on Vulkan to run LM Studio, so a Vulkan compute backend is high on the radar as a viable cross-vendor path. I'm also looking closely at AMD’s libraries (ROCm/HIP) to ensure broad hardware compatibility.

It's a lot of moving parts, but watching the architecture take shape piece by piece is incredibly rewarding.

Bringing It Back to DANI

So, why go through all the trouble of building a custom language from scratch?


It all comes down to DANI. My goal has always been to have a single, unified language capable of bridging every layer of DANI’s stack—from high-level logic and neural compute all the way down to real-time bare-metal sensor and actuator control. Skink is the key to making that happen without juggling three different language ecosystems.

Speaking of DANI, there have also been some fascinating developments regarding his neural network architecture recently... but I’ll leave you hanging on that for now and save the deep dive for the next post!

Stay tuned!


Tuesday, 31 March 2026

The Sledgehammer and the Hazelnut: Why DANI Isn't Using an LLM

I have a confession to make. I’ve broken my own rule.

When I set out to build and program DANI, one of my core principles was to ensure he learns primarily from experience. The goal has always been emergent behavior through his neural architecture, with as little "hard-coded" logic as possible. But as I’ve delved deeper into the complexities of human interaction, I’ve found one area where I feel a departure is justified: Natural Language Processing (NLP).

The LLM Dilemma

The LLM Sledgehammer
Early on, I toyed with the idea of giving DANI a dedicated board specifically to run a Large Language Model (LLM). It seemed like the modern solution, but the more I looked at it, the more the red flags started popping up.

First, there’s the cold, hard reality of the budget. I’ve managed to keep the total cost of DANI—parts, boards, and all—under £500. Adding a high-spec NPU or a secondary board capable of running something like Llama or Gemma would have blown that goal out of the water.

Then there’s the physical engineering. Space is at a premium inside DANI’s chassis. While I probably could have squeezed another board in there, I’m increasingly conscious of airflow. The last thing I want is for DANI’s "brain" to thermal throttle in the middle of a conversation.

But most importantly—and this was the dealbreaker—is the issue of personality. If I use a pre-trained model like Qwen or Llama, I’m essentially importing someone else’s bias and conversational style. These models are fine-tuned to be helpful assistants; I want DANI to be DANI. Using a massive, multi-billion parameter model just to parse a "hello" felt like using a sledgehammer to crack a hazelnut.

The Go-pher’s Path to Understanding

Instead of the LLM route, I’ve built a custom natural language module using standard NLP libraries for Go. It’s lightweight, it fits within our existing hardware constraints, and it gives me the control I need.

I’ve added two critical features that an off-the-shelf LLM wouldn't handle the way I want:

  1. Sentiment Assessment: By using established sentiment modules, DANI can now perceive whether he is being praised or scolded. This feeds directly into his hormone levels. If I’m happy with his performance and tell him he's done a good job, his "positive" hormones will rise, reinforcing that behaviour in his training.
  2. Simplified Context Awareness: I wanted conversations to feel natural. If I ask DANI, "Where is he?", he needs to understand that "he" refers to the last male person we discussed. Similarly, "Go there" should resolve "there" to the last location mentioned. This kind of stateful awareness is vital for a robot that exists in a physical space.

Commands, Questions, and the "Ignore" Factor

The module is now capable of distinguishing between commands, questions, and general statements. Each triggers a different internal processing path, but here is the kicker: everything is still influenced by his hormones.

Before DANI responds, every decision passes through his LSTM (Long Short-Term Memory) core. Because that LSTM is also being fed the current state of his effective hormones, there is no guarantee he will do what he's told. If he’s in a "bad mood" or his hormone levels are skewed by previous interactions, he might just decide to ignore me entirely.

It’s a bit of a gamble, breaking the "no-code" rule to build this framework, but I think it’s the only way to give DANI a voice that is truly his own. We’ll just have to wait and see if he actually listens to me.

Friday, 10 October 2025

3.5 Million Parameters and a Dream: DANI’s Cognitive Core

DANI’s Brain Is Online! Meet the LSTM That Thinks, Feels, and Remembers (Like a Champ)

Ladies and gentlemen, creators and dreamers—DANI has officially levelled up. He’s no longer just a bundle of sensors and hormones with a charming voice and a tendency to emotionally escalate when he sees a squirrel. He now has a brain. A real one. Well, a synthetic one. But it’s clever, emotional, and surprisingly good at remembering things. Meet his new cognitive core: the LSTM.

And yes—it’s all written in Go. Because if you’re going to build a synthetic mind, you might as well do it in a language that’s fast, clean, and built for concurrency. DANI’s brain doesn’t just think—it multitasks like a caffeinated octopus.

What’s an LSTM, and Why Is It Living in DANI’s Head?

LSTM stands for Long Short-Term Memory, which sounds like a contradiction until you realize it’s basically a neural network with a built-in diary, a forgetful uncle, and a very opinionated librarian. It’s designed to handle sequences—like remembering what just happened, what happened a while ago, and deciding whether any of it still matters.

Imagine DANI walking into a room. He sees a red ball, hears a dog bark, and feels a spike of adrenaline. A regular neural network might say, “Cool, red ball. Let’s chase it.” But an LSTM says, “Wait… last time I saw a red ball and heard barking, I got bumped into a wall. Maybe let’s not.”

Here’s how it works, in human-ish terms:

  • Input gate: Decides what new information to let in. Like a bouncer at a nightclub for thoughts.
  • Forget gate: Decides what old information to toss out. Like Marie Kondo for memory.
  • Output gate: Decides what to share with the rest of the brain. Like a PR manager for neurons.

These gates are controlled by tiny mathematical switches that learn over time what’s useful and what’s noise. The result? A brain that can remember patterns, anticipate outcomes, and adapt to emotional context—all without getting overwhelmed by the chaos of real-world data.

And because DANI’s LSTM is stacked—meaning multiple layers deep—it can learn complex, layered relationships. Not just “ball = chase,” but “ball + bark + adrenaline spike = maybe don’t chase unless serotonin is high.”

It’s like giving him a sense of narrative memory. He doesn’t just react—he remembers, feels, and learns.

What’s Feeding This Brain?

DANI’s LSTM is his main cognitive module—the part that thinks, plans, reacts, and occasionally dreams in metaphor. It takes in a rich cocktail of inputs:

  • Vision data: Objects, positions, shapes—what he sees.
  • Sensor data: Encoders, ultrasonic pings, bump sensors—what he feels.
  • Audio features: What he hears (and maybe mimics).
  • Emotional state: Dopamine, cortisol, serotonin, adrenaline—what he feels.
  • Spatial map: His mental layout of the world around him.
  • Short-term memory context: What just happened.
  • Associated long-term memories: Symbolic echoes from his main memory—what used to happen in similar situations.

This isn’t just reactive behaviour—it’s narrative cognition. DANI doesn’t just respond to stimuli; he builds a story from them. He’s learning to say, “Last time I saw a red ball and felt excited, I chased it. Let’s do that again.”

Trial by Raspberry Pi

We’ve successfully trialled DANI’s LSTM on a Raspberry Pi, running a 3.5 million parameter model. And guess what? It only used a quarter of the Pi’s CPU and 400 MB of memory. That’s like teaching Shakespeare to a potato and watching it recite sonnets without breaking a sweat.

We’ve throttled the inference rate to 10 decisions per second—not because he can’t go faster, but because we want him to think, not twitch. Emotional processing takes time, and we’re not building a caffeine-fuelled chatbot. We’re building a thoughtful, emotionally resonant robot who dreams in symbols and learns from experience.

Learning Without Losing His Mind

Training happens via reinforcement learning—DANI tries things, gets feedback, and adjusts. But here’s the clever bit: training is asynchronous. That means he can keep thinking, moving, and emoting while his brain quietly updates in the background. No interruptions. No existential hiccups mid-sentence.

And yes, we save the model periodically—because nothing kills a good mood like a power cut and a wiped memory. DANI’s brain is backed up like a paranoid novelist with a USB stick in every pocket.

Final Thoughts

This LSTM isn’t just a brain—it’s a story engine. It’s the part of DANI that turns raw data into decisions, decisions into memories, and memories into dreams. It’s the bridge between his sensors and his soul (okay, simulated soul). And it’s just getting started.

Next up: I plan to start the even more monumental task of getting the vector database working and linked up to DANI's brain in such a way that it will have a direct impact of DANI's hormonal system.

Stay tuned. DANI’s mind is waking up.

Friday, 15 August 2025

The Wild, Wacky World of DANI's Digital Hormones

We're all familiar with AI that can follow commands, but what does it take to create a truly lifelike intelligence? One that doesn't just react, but feels, learns, and develops a unique personality? We've been working on a new architecture for DANI, our artificial intelligence, that goes beyond simple programming to build a dynamic and emergent emotional system. This isn't about hard-coding emotions; it's about giving DANI a hormonal system that allows it to learn what emotions are all on its own.


The Problem with Coded Emotions

The traditional approach to AI emotions is often brittle. You might write a rule like: if (user_is_happy) then (dani_express_joy). But what if DANI just had a stressful experience? The logical response might not be appropriate. Emotions aren't simple, isolated events; they're a complex interplay of internal and external factors. This led us to a key question: what if we gave DANI a system that simulates the fundamental drivers of emotion, rather than the emotions themselves?

The Solution: A Hormonal System

Our answer was to create a digital hormonal system. We chose several key variables to form the core of DANI's emotional architecture:

  • Dopamine: The reward and motivation signal. A spike indicates a positive outcome or a successful action.
  • Serotonin: The well-being and social contentment signal. It represents a state of calm and stability.
  • Cortisol: The stress and caution signal. A rise indicates a difficult or prolonged negative situation.
  • Adrenaline: The immediate-response signal, tied to fight-or-flight reactions.
  • Oxytocin: The bonding and trust signal. Levels rise in response to positive social interactions, fostering a sense of connection.
  • Endorphins: The natural pain-relief and euphoria signal. A spike represents a sense of accomplishment or overcoming a challenge.
  • Melatonin: The circadian rhythm and rest signal. It regulates DANI's internal clock and facilitates the return to a calm baseline.

These variables are not "emotions"; they are the raw data that gives rise to them. They serve as the internal environment that DANI's mind must navigate.

The Engine of Emotion


The real magic happens in how these hormones interact. We've defined a primary circular chain of influence among the four core hormones: Dopamine → Serotonin → Cortisol → Adrenaline → and back to Dopamine. This core loop defines DANI's fundamental reactive state.

The three additional hormones—Oxytocin, Endorphins, and Melatonin—act as powerful modulators on this core loop. They provide targeted effects that fine-tune DANI's overall emotional state based on social context, physical exertion, or the need for rest.

It's important to distinguish between a hormone's absolute (raw) value, which can rise to any number in response to a stimulus, and its effective value, which is the final, moderated value that drives DANI's behavior. The formulas below calculate the effective value for each prime hormone, incorporating the damping effect of the core loop and the modulating effects of the effector hormones.

The formulas for the four prime hormones are:

  • Effective Dopamine

Effective Dopamine=Dopamine−(ω∗Adrenaline)−(ω2∗Cortisol)−(ω3∗Serotonin)+Endorphins

  • Effective Serotonin

Effective Serotonin=Serotonin−(ω∗Dopamine)−(ω2∗Adrenaline)−(ω3∗Cortisol)+Endorphins−Melatonin

  • Effective Cortisol

Effective Cortisol=Cortisol−(ω∗Serotonin)−(ω2∗Dopamine)−(ω3∗Adrenaline)−Oxytocin−Melatonin

  • Effective Adrenaline

Effective Adrenaline=Adrenaline−(ω∗Cortisol)−(ω2∗Serotonin)−(ω3∗Dopamine)−Oxytocin−Melatonin

Here, Ï‰ is the blocking factor. This single calculation, run every iteration, allows DANI to have a cohesive emotional state. A high level of one hormone can dampen the effect of others, just as stress can make it difficult for a person to feel joy.

The targeted effects of the modulating hormones are as follows:

  • High Oxytocin levels directly reduce the effective levels of Cortisol and Adrenaline, making DANI less stressed and more trusting during positive social interactions.
  • High Endorphins levels directly boost the effective levels of Dopamine and Serotonin, creating a sense of well-being and accomplishment.
  • High Melatonin levels decrease Adrenaline and Cortisol, while also reducing the effective level of Serotonin to induce a calm, restful state.

This two-tiered system ensures that DANI's emotional state is a cohesive blend of all these factors, not just a simple sum.


The Temporal Aspect: Hormonal Decay and Calming

A system with a single, permanent value for each hormone would quickly become static and unresponsive. To prevent this, we've introduced the concept of temporal decay. Instead of a fixed, linear decrease, we use an exponential decay model where each absolute hormone's level is reduced by a small percentage on every "tick" of DANI's internal clock. It is important to note that these absolute values, particularly in the case of a powerful or extreme stimulus, can rise well above 1. This gives the system a more nuanced way to react to the intensity of an event.

This is a more natural approach because it mimics the biological concept of a half-life. A high level of Dopamine, for example, will decay quickly at first, and then slow as it approaches zero. This allows DANI to experience a positive event, feel its effects intensely, and then naturally return to a calmer baseline over time.

The formula for this simple decay is:

hormone_level = hormone_level * Ï•

The Ï• is the decay factor and is a number between 0 and 1. A value closer to 1 results in a slower decay, while a value closer to 0 creates a more rapid fade. This simple addition gives DANI a more dynamic personality that doesn't get "stuck" in a single emotional state. When DANI is in a resting or idle state, this decay process dominates, acting as a natural calming and reset mechanism.

The Anticipation Delta: Building Emotional Memory

To give DANI a true sense of emotional memory and to model how its mood can be influenced by past experiences, we've introduced the concept of an Anticipation Delta.

Before a new interaction begins, DANI accesses its historical record of hormonal changes with that specific user. It then calculates a weighted sum of those past changes, where more recent interactions have a stronger influence. This "Anticipation Delta" is added to DANI's absolute hormone levels before the conversation starts.

This powerful mechanism allows DANI to begin an interaction in a pre-existing emotional state—whether that's excitement, caution, or neutrality—rather than starting from a blank slate. Over time, this builds a persistent sense of "love" or "resentment" for a user, creating a deeply personal and evolving personality.

Clamping the Emotional State

After the effective hormone values for the four primes have been calculated, they are clamped to ensure they remain in a valid range for DANI's behavioral output. Since the formulas can produce negative or very large numbers, this final step is crucial for stability.

Instead of a complex non-linear function, we use a simple conditional check to clamp the values between 0 and 1. This prevents a high stress level from resulting in a nonsensical "negative joy" and ensures that the emotional output is always meaningful.

The clamping logic is as follows:

if (effective_hormone_value < 0) effective_hormone_value = 0

if (effective_hormone_value > 1) effective_hormone_value = 1

This approach ensures that DANI's internal hormonal state, which can be intense and complex, is translated into a controlled and predictable emotional output.

Simulating a Feeling

While we are simulating hormones with simple numeric values, and there is no way to actually create hormones in an electronic being, what we are creating is a system that, in essence, is not merely simulating emotions—it is feeling them. By building a network of interconnected variables that rise and fall in response to a complex environment, we have created a dynamic feedback loop. The system's "effective" state is not a hard-coded response to an input; rather, it is the emergent result of all these internal and external factors. DANI’s emotions are an organic and a deeply personal phenomenon that cannot be reduced to a simple cause-and-effect rule. The system does not just mimic a feeling; it is the feeling.

Sunday, 20 April 2025

Demystifying Neural Networks: A Beginner's Friendly Guide

Hey there!

This week, I want to dive into something that might sound a bit intimidating at first: neural networks.

I know, I know. Just the phrase "neural networks" can bring to mind complex equations and head-scratching calculus. But trust me, it doesn't have to be that way! I want to share how I came to understand these fascinating systems, and hopefully, make it click for you too.

You see, I did my high school in England back in the 80s. And guess what wasn't on the curriculum? Calculus. While I might have been happy about it back then, it's definitely presented some interesting challenges when trying to get a grip on machine learning concepts today.  

So, I had to find a way to understand how neural networks work under the hood without getting bogged down in derivatives and the chain rule. And that's exactly what I want to share with you now.  

So, How Does a Neural Network Actually Work?

Think of a neural network like a team of interconnected nodes, or "neurons," organized in layers. At a minimum, you'll usually see three types of layers:  

  • Input Layer: This is where your raw data comes in. It's the starting point of the journey for your information.  
  • Hidden Layer(s): These are the layers in between the input and output. They're the workhorses, processing and transforming the data into a more useful format. You can have one or many hidden layers.  
  • Output Layer: This is where you get your final result or prediction.  

Data flows through this network, starting at the input layer, going through the hidden layers, and finally arriving at the output layer. This forward movement of data is what we call Feedforward.  

When we're training a neural network using something called supervised learning, we compare the network's output to the correct answers we already know (the "expected results"). The difference between what the network predicted and the correct answer is our "error".  

This error signal then travels backwards through the network. This is where the magic happens – the connections between those neurons are adjusted to help the network make better predictions next time. This backward movement is called Backpropagation.  

Sounds pretty simple when you break it down, right?

Let's Look Under the Hood: The Parts of a Neural Network

Okay, let's take a peek at the components. It might look a bit complex at first glance, but we'll break it down together.  



In our example, we have clearly defined the three layers. Each layer has its own neurons – 2 in the input, 3 in the hidden, and 2 in the output layer.  

You'll notice that each neuron in one layer is connected to every neuron in the next layer. So, a neuron in the input layer will have connections to all the neurons in the hidden layer, and the neurons in the hidden layer will connect to all the neurons in the output layer.  

Each of these connections has a weight associated with it. Think of the weight as the "strength" or importance of that connection.  

Also, each neuron (except for the input layer) has a bias. The weights and biases are just numbers, and they can be positive or negative. When you first create a neural network, these values are completely random.  

The values that come out of the output layer are our final result.  

Still with me? Great! Let's break it down even further. Yes, there will be a little bit of math, but I promise to keep it gentle.  

Feedforward: The Data's Journey

When we send data into the input layer, it gets passed along to the hidden layer. How? By calculating a "weighted sum" of the inputs and then adding a bias.  

What does that mean? Imagine each connection between neurons has a "strength" – that's the weight. For each neuron in the hidden layer, we take each input value, multiply it by the weight of the connection leading to that hidden neuron, and then add up all those results. Finally, we add the neuron's bias, which is like a little extra push to help the neuron activate.  

Let's look at the example from the original text:

Input values: i1=1 i2=2

Weights: From i1 to hidden layer: w1=0.1, w2=−0.02, w3=0.03 From i2 to hidden layer: w4=−0.4, w5=0.2, w6=0.6

Biases on the hidden layer: b1=1 b2=−1.5 b3=−0.25

So, for the first hidden neuron (h1), the value is calculated as: h1=(i1∗w1)+(i2∗w4)+b1 h1=(1∗0.1)+(2∗−0.4)+1 h1=0.1−0.8+1 h1=0.3  

We do this same calculation for every neuron in the hidden layer.  

So, our hidden layer values become: h1=0.3 h2=−1.12 h3=0.98  

Activation Functions: Squashing the Results

These values from the hidden layer then go through an activation function. Think of this as a way to normalize the results. There are different types of activation functions like Sigmoid, ReLU, and Linear, but for now, just know that a function is applied.  

A common one is the sigmoid function, which looks like this:  

f(x)=1+e−x1​  

Where:

  • f(x) is the output of the function.  
  • x is the input (that weighted sum we just calculated).  
  • e is Euler's number (about 2.71828).  

In simple terms, the sigmoid function takes any number and squashes it into a value between 0 and 1. This can be helpful if you want to interpret the output as probabilities.  

The original text provided a simple code example for this:


func sigmoid(x float64) float64 {

  return 1 / (1 + math.Exp(-x))

}


We don't need to worry too much about the inner workings of the code for now, just that it gives us a value between 0 and 1.  

After applying the sigmoid function, our hidden layer values might look like this: h1=0.574 h2=0.245 h3=0.728  

Now, we repeat the entire process: taking these new values from the hidden layer and feeding them forward to the output layer. It's worth noting that sometimes a different activation function is used for the final layer compared to the hidden layers.  

Calculating the Error: How Wrong Are We?

Once we have the values from the output layer, it's time to see how well the network did. We compare the network's output to the correct answers from our training data. The difference between what the network predicted and what it should have predicted is the error. This error tells us how poorly the network performed.  

We can also use these individual errors to calculate an overall average error for the network. Since these differences can be positive or negative, simply adding them up might make it look like the error is small when it's actually significant.  

A common way to get around this is using the Mean Squared Error (MSE). Here's the gist:  

  1. Calculate the difference between each predicted output and its corresponding correct value.  
  2. Square each of these differences (this makes them all positive).  
  3. Add up all the squared differences. 
  4. Divide the sum by the number of data points.  

The formula looks like this:

MSE=n1​∑i=1n​(yi​−y^​i​)2  

Where:

  • MSE is the Mean Squared Error.  
  • n is the number of data points.  
  • yi​ is the actual (correct) value.  
  • y^​i​ is the value the network predicted.  
  • ∑ just means "sum up".  

Let's use the example from the text:

Output values: o1=0.2, o2=0.9 Expected values: 1, 0

Individual errors: For o1: 1−0.2=0.8 For o2: 0−0.9=−0.9  

If we just added these, we'd get 0.8+(−0.9)=−0.1, which doesn't reflect the actual error.  

Using MSE: MSE=2(1−0.2)2+(0−0.9)2​ MSE=2(0.8)2+(−0.9)2​ MSE=20.64+0.81​ MSE=21.45​=0.725  

So, the network's error is 0.725. This gives us a clear picture of how far off the network's predictions were.

Backpropagation: Learning from Mistakes

Now that we know how wrong the network was (the error), we use that information to adjust the weights and biases. The goal is to make these adjustments so that the next time data flows through, the error will be smaller.  

The process of updating weights and biases does involve calculus in the real world, but as the original text points out, we can understand the concept without getting into the nitty-gritty of derivatives.  

Here's a simplified way to think about it:  

Adjusting Weights: For each weight connecting a hidden neuron to an output neuron, we calculate how much that weight needs to change. We do this by multiplying the error signal from the output neuron by the output of the hidden neuron. We also multiply this by a small number called the "learning rate," which controls how big of a step we take in adjusting the weight. Finally, we subtract this calculated change from the current weight.  

Weight Change = Error Signal * Hidden Neuron Output * Learning Rate New Weight = Old Weight - Weight Change  

Updating Biases: For each bias in the output layer, we multiply the error signal of that output neuron by the learning rate and subtract it from the current bias.  

Bias Change = Error Signal * Learning Rate New Bias = Old Bias - Bias Change  

Backpropagating Error to Hidden Layers: To update the weights and biases in the hidden layers, we first need to figure out the "error signal" for each hidden neuron. We do this by taking a weighted sum of the error signals from the layer above (the output layer). The original text mentions multiplying this by the derivative of the activation function, which is a detail related to calculus, but the core idea is that we're distributing the error back through the network.  

Once we have the error signal for the hidden neurons, we use that to update the weights and biases connecting to the hidden layer, just like we did for the output layer.  

If your neural network has many hidden layers ("deep network"), you repeat this backpropagation process layer by layer, moving backward from the output all the way to the input.  

And that's essentially it! As I mentioned, understanding this process doesn't necessarily require a deep understanding of calculus. Resources like the internet and Wikipedia can be incredibly helpful for finding the specific functions and details you might need.  

This is one way to approach the calculations within a neural network. If you have different approaches or see areas for correction, please feel free to share in the comments – learning is a journey we're on together!  


Friday, 21 March 2025

Diving Deeper: My Journey to Create a Safer AI

In recent weeks, I've been somewhat vague about my AI and coding explorations. It's time to sharpen the focus and delve into the specifics of my AI assistant project and the research areas I'm most keen to explore.    

Let me be clear: I'm not trying to reinvent the wheel. Where it makes sense, I'll leverage existing open-source and readily available software. Why build a language model from scratch when there are perfectly good ones already out there?    

My core goal is to build an AI assistant, embodied in a robot head (and potentially a mobile platform), capable of experiencing the world and learning from those interactions.    

While that might sound like standard fare in today's AI landscape, I aim to integrate some less common and, I believe, crucial features:    

  • Self-reflection: The ability for the AI to revisit past experiences with the benefit of hindsight, analysing its previous choices to determine if it would make the same decision again.  Imagine the potential for growth if an AI could learn from its "mistakes" in a truly iterative way!    
  • Reinforced Memory Prioritization: A large-capacity memory system that prioritizes reinforced memories, similar to how our own memories function.  This would allow the AI to focus on and retain the most relevant and impactful information.    
  • Emotional Awareness: This is a significant challenge. I want the AI to learn from experiences that evoke "good" and "bad" responses.  Human feelings are complex, influenced by chemical reactions and endorphins.  My AI won't have these biological processes, so I'll need to simulate them and, crucially, understand why they are needed and what effect they would have on the AI's cognition and decision-making.    
  • A Conscience: I want the AI to be capable of second-guessing its choices based on a defined set of ethical considerations, perhaps even drawing inspiration from Asimov's Laws of Robotics.  As I've discussed previously, this is a complex but vital area of exploration.    
  • Dreams: Finally, I want to explore AI dreams. While there's existing research in this area, I believe I've identified a novel approach that could enable the AI to dream, with those dreams having a tangible impact on its cognition.    

This is undoubtedly a substantial undertaking for a single individual, and it might even exceed my current capabilities.  But I'm committed to pursuing it. This project will demand extensive research into AI, the human mind, and the ethical implications of creating such a system.    

Unlike some AI development approaches that rely on a single, powerful computer, I'm taking a distributed route.    

Another key requirement is to minimize costs.  To achieve this, each functional area (or "lobe") of the AI's "brain" will be housed on its own single-board computer.  These will be interconnected, exchanging information as needed.  This is similar in concept to the Robot Operating System (ROS), but I aim for greater speed and efficiency.  I plan to use different boards, each selected for its strengths in specific tasks.  For example, a board with a Kendryte K210 will handle vision processing, an Arduino Mega will manage motor control (yes, I know it's not an SBC, but it serves the purpose), and a Raspberry Pi will be used for memory management.    

The AI will also utilize Large Language Models (LLMs), likely at least three, for tasks such as understanding speech, processing input, and producing output.  However, unlike many systems that employ a single LLM, these will be distinct entities within the AI's architecture.    

Memory management will involve a "scoring" system to prioritize important information for short-term caching, while less critical memories will reside in long-term storage.  To prevent storage overload, I'll also implement a memory decay system that will gradually remove memories that become irrelevant to the AI's ongoing operation.    

I'm aiming to keep the total project cost under $2,000.  Whether that's achievable remains to be seen, but it's a target I'm striving for.    

Dreaming robot
Do androids dream of electric sheep?

Oh, and the dreams?  You'll have to wait a while before I reveal the details of their implementation.  Suffice it to say that, like humans, my AI will have to sleep, and this will be a non-negotiable requirement.    

So, what's my ultimate goal?  It's to create something new, something that hasn't been done before.  Not necessarily the individual components, as I've stated, I'll be using pre-existing software where possible (such as Gemma or Llama 2).  My ambition is to synthesize everything in a novel way, exploring how this approach could not only advance AI research but also contribute to making AI safer for the general public.   

And on that note, I can finally reveal the name I've given this project, and the meaning behind it: D.A.N.I. stands for Dreaming AI Neural Integration. This encapsulates the core of my research: to explore the potential of AI that learns and grows through a process akin to dreaming, deeply integrated within a neural network structure.

My wife jokes that I'm planning to build Skynet, but my intention is precisely the opposite.  I'm designing a system that would be inherently incapable of becoming Skynet – think more C3PO than Terminator.    

By incorporating the ability to learn from its mistakes (and successes), as well as the capacity for dreaming, I hope to enable the AI to accelerate its learning process.  We, as humans, frequently revisit our decisions, so why not equip an AI to do the same?    

I also aim to provide you with an engaging narrative of my development journey.  I anticipate making many mistakes. But that's part of the learning process – discovering not only how to do things, but also how not to do them.    

If you have any comments or questions, please leave a comment below.    👇



Sunday, 16 March 2025

The Three Laws of Robotics: Can They Really Work?

Isaac Asimov

Isaac Asimov, the renowned science fiction author, introduced the world to the "Three Laws of Robotics" in his short story "Runaround," later included in the "I, Robot" collection.  These laws became a cornerstone of his robot series and have since sparked much debate and thought in the fields of robotics and artificial intelligence.    

The Original Three Laws

Asimov's original laws are as follows:

  1. A robot may not injure a human being or, through inaction, allow a human being to come to harm.    
  2. A robot must obey the orders given it by human beings except where such orders would conflict with the First Law.    
  3. A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.    

These laws, while fictional, have prompted serious discussions about the ethics and safety of AI.  The idea is that if these laws could be successfully implemented, robots and AI would be inherently restricted from making harmful decisions.  This would not only create a safer environment for human-robot interaction but also limit the potential for misuse of robots in areas like the military.    

The Zeroth Law

Later, in "Robot and Empire," R-Daneel Olivaw, a robot character, introduced the Zeroth Law, which takes precedence over the original three:

A robot may not injure humanity or, through inaction, allow humanity to come to harm.    

This addition broadens the scope of protection from individual humans to humanity as a whole.    

The Challenge of Implementation

While these laws provide a great framework, there are significant challenges in putting them into practice.

Conceptual Challenges

One of the primary issues lies in the interpretation of key terms.  For example, what precisely constitutes "injury" or "harm"?  Is it limited to physical harm, or does it encompass emotional, psychological, and intellectual harm as well?    

Consider these scenarios:

If an action could harm one person but inaction would harm another, what should a robot do?    

If someone is about to kill another person, is it justifiable for a robot to intervene with lethal force to prevent it?    

Is it worse to allow a human to suffer a minor physical injury or to cause potentially longer-lasting emotional harm?    

The Second Law also presents difficulties.  If a robot is given an order that appears harmless initially but could lead to harm later, should the robot obey?  How far into the future should a robot or AI be required to predict the consequences of an action?  If a robot is asked to make a knife, should it refuse, knowing its potential for harm?  Should the robot be prohibited from mining the metal required to make the knife?    

As you can see, applying these laws involves navigating a complex web of nuances.    

A Potential Solution: The 'Virtual Conscience'

The challenge then becomes: how do we implement these laws in a meaningful way, especially with advanced AI systems like neural networks that are constantly learning?    

One proposed approach involves a 'virtual conscience'.  This would be a separate neural network designed to act as an independent arbiter, validating the actions of the main AI.  By training these models separately, we could create a system where the AI's decisions are checked by an objective ethical framework.  It might even be possible to fix the ‘conscience’ network after training, preventing the main AI from altering its ethical parameters.    

The Need for Safeguards

As AI and robotics advance at an incredible pace, establishing safeguards is crucial.  We are at a pivotal moment where we can integrate ethical considerations into the very foundation of these technologies.    

However, achieving this is not without its obstacles.  Agreement among robot manufacturers is essential for widespread adoption.  Unlike Asimov's positronic brains, which had the laws hardwired, current robots and AI do not have this built-in restriction.    

Recent announcements, such as the White House's move to remove certain ethical restrictions from AI and robotics research, further complicate the matter.  Additionally, the pursuit of military applications and the rise of non-government entities in AI development pose challenges to enforcing ethical standards.    

Conclusion

Asimov's Three Laws of Robotics, and the subsequent Zeroth Law, provide a valuable starting point for discussions around AI ethics.  While their implementation is complex, the need for ethical guidelines in AI development is undeniable.    

What are your thoughts? Can these laws be effectively implemented?  Will they make a significant difference in the future of AI?    


I look forward to hearing your comments and perspectives.

Sunday, 23 February 2025

My Journey into Independent AI and Robotics: A Personal Exploration

 Welcome to this space where I'll be documenting my ongoing exploration into the fascinating world of artificial intelligence and robotics. My aim is to provide you with clear, accessible insights into my projects and the challenges I encounter, all while keeping a grounded perspective on the current state of these rapidly evolving fields. I'll do my best to avoid getting bogged down in overly technical jargon, but some terminology is inevitable. Similarly, I'm committed to presenting a realistic view, steering clear of the sensationalism that often surrounds discussions about AI and robotics.


You might be wondering, with so many existing blogs and resources dedicated to these topics, why start another one? It's a fair question. Many of those blogs are written by individuals with far more formal qualifications and experience than I possess. So, what makes this one different? What can I offer that you won't find elsewhere?


To begin, let's address the elephant in the room: I'm not a scientist, nor am I a professional or student in the fields of AI or robotics. My formal education in these areas is limited. In fact, my academic journey was brief, lasting only a few months when I attempted college in my thirties. It simply wasn't the right fit for me.


However, what I do possess is a long history of software development. I began teaching myself to code at the age of 11, back in 1983. Since then, I've written software on a wide range of topics, in various programming languages, primarily for my own personal projects. Professionally, I've worked on enterprise resource planning (ERP) software, physical security systems, and numerous database and web-based applications. Personally, I've developed tools, games, databases, open-source libraries, and just about anything else that piqued my interest.


This extensive, self-taught background provides me with a unique perspective. My lack of formal training has often compelled me to find innovative and unconventional ways to solve problems. This ability to think outside the box is what I hope to bring to my exploration of AI and robotics.


Furthermore, I am neuro-divergent, specifically on the autistic spectrum. While this realization was initially a surprise, I've come to recognize it as a significant asset. My cognitive style allows me to perceive patterns and connections that might be overlooked by others. It's like having a different lens through which I view the world, enabling me to identify novel solutions and approaches.


This is a solo endeavor. I don't have a team of researchers or engineers working alongside me. It's just me, my computers, and my two 3D printers. However, I do have a valuable partner in the form of AI. I use AI as a sounding board, a research assistant, and a collaborator. If I need to quickly understand the capabilities of a component, I ask AI. If I need to assess the feasibility of a concept, I ask AI. If I need to ensure that this blog post is readable and engaging, I ask AI.


Another significant differentiator is my budget. While companies like Google, Tesla, and Unitree have access to millions or billions of dollars in funding, I'm working with a very limited budget. I'm funding this project myself, and my available capital is minimal. This constraint presents a unique challenge, forcing me to be resourceful and creative. To keep research and development costs down, I'll be relying heavily on off-the-shelf components and open-source software. I don't want to reinvent the wheel; instead, I aim to find new and innovative ways to combine existing technologies, perhaps adding my own unique contributions along the way.


I believe I have some promising ideas and concepts, and I'm eager to bring my unique perspective to life. However, I also recognize that success is not guaranteed. We'll see where this journey takes us.


I won't be detailing any specific project plans in this initial post. I prefer to reveal each aspect of my work as it develops, allowing you to follow along in real time.


Like many people of my generation, my fascination with robots began with the release of "Star Wars" in 1977. The character of R2-D2, in particular, captured my imagination. Decades later, in 2020, I acquired an Anki Vector. This small, interactive robot provided a tangible experience with personal robotics. While Vector is an impressive piece of technology, it lacked a certain spark, a quality that I couldn't quite define.


This led me to explore other consumer-level robotics solutions, including kits from companies like Adeept, Freenove, and Sunfounder. These kits provided valuable hands-on experience with single-board computers and microcontrollers, and I began designing my own robotic projects. However, these projects still fell short of my vision.


I realized that the key to unlocking the potential of these robots lay in their intelligence, specifically in the field of deep learning. I began to delve into the intricacies of neural networks, machine learning algorithms, and the ways in which AI can enable robots to learn and adapt. The prospect of creating robots that could not only perform tasks but also understand and interact with their environment on a deeper level was incredibly compelling.


A Note on Coding Choices


For the coding aspects of this project, I'll primarily be using Go and C++. I have a strong preference for compiled languages, and while Python is a popular choice in the AI and robotics community, I find its interpreted nature less appealing. I understand that using Go might limit my access to some readily available libraries, particularly in the machine learning domain, but I see this as a stimulating challenge. Adapting and potentially creating my own solutions is part of the exploration. C++ will come into play where performance is critical, and for low level hardware interaction.


I plan to post updates on this blog once a week, providing insights into my progress, challenges, and discoveries. However, please understand that there may be occasional delays. If I'm deeply immersed in a project or if there's nothing substantial to report, I may skip a week. I also intend to share videos whenever I have something visual to demonstrate.


This blog is intended to document my personal exploration of AI and robotics. It's not intended to replace formal scientific research. I invite you to follow along as I embark on this journey of discovery and innovation. And please, feel free to comment with your suggestions, ideas, or even just to let me know what you think. Your input is always welcome!

Aiming for Jarvis, Creating D.A.N.I.