BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Really?
So are you saying that all those techinal persons can't see that and here you come and solve all their problems huh?
It's aperently a bit more complicated that that.
124000 people (minus 15000) and I am pretty sure a good deal of them are smarter than you, sorry to be the bearer of bad news.
Pom down under
Top 20
Intel either embrace us or end up on the scrap heap selling themselves out for a few $$$ Peter/Anil oops sorry I mean @Diogenese
PVDM is one person. But you have truly just written a BS post. Thats because of who you are as a person. You need to do better. Much better.
Last edited:
Really, what's wrong of being realistisk huh?
Go to bed Bacon
Totally agree with ur opinion.
It never even crossed my mind that you were suggesting a takeover Dio..I created that all by myself
FromBeyond
Member
Worm alert. It's shovel time.
Fullmoonfever
Top 20
Looks like some friends in Japan, with a little support from Megachips, have been playing with Akida & MetaTF 
Apols if already posted as I may have missed it and haven't done a search.
Short video end of post.
Paper HERE
License: arXiv.org perpetual non-exclusive license
arXiv:2408.13018v1 [cs.RO] 23 Aug 2024
Excerpts:
Table 3: Hardware performance of policies: FPNN was evaluated by edge-CPU (Raspberry Pi 4: quad-core ARM Cortex-A72). SNN was evaluated by neurochip (Akida 1000 [9]). “Power cons” and “Calc. speed” denote power consumption and calculation speed for obtaining one action from NN policies using each piece of hardware. Power consumption was measured by voltage checker (TAP-TST8N).
Apols if already posted as I may have missed it and haven't done a search.
Short video end of post.
Paper HERE
License: arXiv.org perpetual non-exclusive license
arXiv:2408.13018v1 [cs.RO] 23 Aug 2024
Robust Iterative Value Conversion: Deep Reinforcement Learning for Neurochip-driven Edge Robots
Yuki Kadokawakadokawa.yuki@naist.ac.jpTomohito Koderakodera.tomohito.kp9@is.naist.jpYoshihisa Tsuruminetsurumine.yoshihisa@is.naist.jpShinya Nishimuranishimura.shinya@megachips.co.jpTakamitsu Matsubaratakam-m@is.naist.jpNara Institute of Science and Technology, 630-0192, Nara, Japan MegaChips Corporation, 532-0003, Osaka, JapanAbstract
A neurochip is a device that reproduces the signal processing mechanisms of brain neurons and calculates Spiking Neural Networks (SNNs) with low power consumption and at high speed. Thus, neurochips are attracting attention from edge robot applications, which suffer from limited battery capacity. This paper aims to achieve deep reinforcement learning (DRL) that acquires SNN policies suitable for neurochip implementation. Since DRL requires a complex function approximation, we focus on conversion techniques from Floating Point NN (FPNN) because it is one of the most feasible SNN techniques. However, DRL requires conversions to SNNs for every policy update to collect the learning samples for a DRL-learning cycle, which updates the FPNN policy and collects the SNN policy samples. Accumulative conversion errors can significantly degrade the performance of the SNN policies. We propose Robust Iterative Value Conversion (RIVC) as a DRL that incorporates conversion error reduction and robustness to conversion errors. To reduce them, FPNN is optimized with the same number of quantization bits as an SNN. The FPNN output is not significantly changed by quantization. To robustify the conversion error, an FPNN policy that is applied with quantization is updated to increase the gap between the probability of selecting the optimal action and other actions. This step prevents unexpected replacements of the policy’s optimal actions. We verified RIVC’s effectiveness on a neurochip-driven robot. The results showed that RIVC consumed 1/15 times less power and increased the calculation speed by five times more than an edge CPU (quad-core ARM Cortex-A72). The previous framework with no countermeasures against conversion errors failed to train the policies. Videos from our experiments are available:Excerpts:
5.1 Construction of Learning System for Experiments
5.1.1 Entire Experiment Settings
This section describes the construction of the proposed framework shown in Fig. 2. We utilized a desktop PC equipped with a GPU (Nvidia RTX3090) for updating the policies and an Akida Neural Processor SoC as a neurochip [9, 12]. The robot was controlled by the policies implemented in the neurochip. SNNs were implemented to the neurochip by a conversion executed by the MetaTF of Akida that converts the software [9, 12]. Samples were collected by the SNN policies in both the simulation tasks and the real-robot tasks since the target task is neurochip-driven robot control. For learning, the GPU updates the policies based on the collected samples in the real-robot environment. Concerning the SNN structure, the quantization of weights 𝑤𝑠 described in Eq. (16) and the calculation accuracy of the activation functions described in Eq. (19) are verified in a range from 2- to 8-bits; they are the implementation constraints of the neurochip [9].Table 3: Hardware performance of policies: FPNN was evaluated by edge-CPU (Raspberry Pi 4: quad-core ARM Cortex-A72). SNN was evaluated by neurochip (Akida 1000 [9]). “Power cons” and “Calc. speed” denote power consumption and calculation speed for obtaining one action from NN policies using each piece of hardware. Power consumption was measured by voltage checker (TAP-TST8N).
| Hardware | Edge-CPU | Neurochip |
|---|---|---|
| Network | FPNN | SNN |
| Power consumption [mW] | 61 | 4 |
| Calculation speed [ms] | 205 | 40 |
7 Conclusion
We proposed RIVC as a novel DRL framework for training SNN policies with a neurochip in real-robot environments. RIVC offers two prominent features: 1) it trains QNN policies, which can be robust for conversion to SNN policies, and 2) it updates the values with GIO, which is robust to the optimal action replacements by conversion to SNN policies. We also implemented RIVC for object-tracking tasks with a neurochip in real-robot environments. Our experiments show that RIVC can train SNN policies by DRL in real-robot environments.Acknowledgments
This work was supported by the MegaChips Corporation. We thank Alonso Ramos Fernandez for his experimental assistance.Great post.
I share the same opinion. Being a top 20 shareholder I am more confidence now than ever with all the deals that we have signed. Looks like this year is definitely our year.
Not advice
Jacob Kiraz
Emerged
How many shares do you need to be a top 20 shareholder
Pom down under
Top 20
Space Cadet
Regular
Seeing as this seems to be a pissing contest with no proof required. I am known to my friends as PVDM.
SC
The Unfolding Drama in the Tech Stocks: Nvidia, BrainChip, and Market Movements
The tech stock market is witnessing dramatic movements, particularly with companies like NVIDIA and BrainChip. Market reactions to earnings reports, AI advancements, and geopolitical factors influe…
heyzul.com
Top 5 Biggest ASX AI Stocks Companies to Invest in 2024 - Globalbiz Outlook
Are you thinking about investing in AI? Find out what technologies the largest pure-play ASX AI stocks are working on and their market capitalization.
globalbizoutlook.com
Top Neuromorphic Computing Stocks for 2025: Ranked by Pure-Play Focus
This guide covers the top neuromorphic computing stocks and companies for investors to watch, ranked by their pure-play focus.
exoswan.com
Smoothsailing
Regular
Extract from a Forbes article talking about LiDAR technology......
Valeo: is designed into the Mercedes S-Class with its SCALA 2 LiDAR (currently in production) to support its L3 autonomy feature. Its next-generation version, the SCALA 3 uses a similar platform as the SCALA 2, but delivers significantly higher range and resolution performance. (SCALA 2 addresses 80 m range and 300,000 pps in a ~ 600 cm³ volume with a square format. SCALA 3 doubles the range and delivers 10M pps in a ~1000 cm³ volume, with a 45 mm height and rectangular format suitable for roof or behind-the-grill mounting). The higher performance of the SCALA 3 is expected to double the speeds at which the L3 autonomy feature can perform. Valeo is also attempting to standardize the SCALA 3 and minimize/eliminate customization to create economies of scale. SCALA 3 is expected to start production by early 2025 and has been selected by Stellantis and another undisclosed global OEM to support L3 autonomy features.
Valeo: is designed into the Mercedes S-Class with its SCALA 2 LiDAR (currently in production) to support its L3 autonomy feature. Its next-generation version, the SCALA 3 uses a similar platform as the SCALA 2, but delivers significantly higher range and resolution performance. (SCALA 2 addresses 80 m range and 300,000 pps in a ~ 600 cm³ volume with a square format. SCALA 3 doubles the range and delivers 10M pps in a ~1000 cm³ volume, with a 45 mm height and rectangular format suitable for roof or behind-the-grill mounting). The higher performance of the SCALA 3 is expected to double the speeds at which the L3 autonomy feature can perform. Valeo is also attempting to standardize the SCALA 3 and minimize/eliminate customization to create economies of scale. SCALA 3 is expected to start production by early 2025 and has been selected by Stellantis and another undisclosed global OEM to support L3 autonomy features.
Not sure if this is a good idea or not, but I’ve just submitted a “send a story” to 60 minutes telling them about the ground breaking tech that could “potentially” address the huge energy problem with current AI. Imagine the positive PR if 60 minutes ran the story??? 
if anyone else would like to also submit, here’s the link https://9now.nine.com.au/60-minutes/send-a-story
if anyone else would like to also submit, here’s the link https://9now.nine.com.au/60-minutes/send-a-storySimilar threads
