Some argue that Brainchip could not possibly have out engineered Intel and the other big players in the neuromorphic engineering space.
Intel’s Loihi 1&2 are held up as AKIDA’s nemesis. Indeed Loihi 2 is often written up in articles as having like AKD1000 on chip learning/training.
Well I was just reading a research paper relating to the use of Loihi 2 and to my surprise given how recent the paper is I found the following:
“B. NEUROMORPHIC MODEL FOR RMM
As seen in Fig. 5, we consider a layered SNN with L = 4 layers, where the hidden layers comprise 512,256 and 512 neurons respectively, and Z = 6. We train the system via the SG-based method SLAYER [46]. SGD is carried out using the Adam optimizer. Models are trained using Intel’s Lava library [47] with Loihi bit-accurate precision, on a single A100 GPU.
On-chip training was not yet available on Intel’s Loihi 2 at the time of writing.
Decisions are obtained via rate decoding, i.e., by selecting the output neuron with the largest spiking rate.
Following the approach proposed in reference [34], train- ing is completed using a dataset D, composed of measure- ments of the required capacity in each geographical zone. Each example R ∈ Rm×n in dataset D consists of n × m resource requirements, in Mbps, for each geographical posi- tion, as detailed in Section II. We preprocess each example independently as follows. First, we set all the outlier values over a given percentile p to the value rp, which is the value such that p% of the entries in R are smaller than rp. Through- out, we set p = 0.98. We then normalize the examples to the range [0, 1], and perform max-pooling with stride ds to reduce the input size to (n/ds) × (m/ds) before encoding into binary spiking signals, as described in Section IV-C.
We perform inference using SNNs as described in the previous sections on Intel’s Loihi 2 chips [47]. Loihi 2 is a research neuromorphic chip that uses asynchronous spiking neurons to implement fine-grained, event-driven, adaptive, self-modifying, parallel computations. Loihi’s first iteration was fabricated on Intel’s 14 nm process and houses 128 clus- ters of 1,024 artificial neurons each, for a total of 131,072 simulated neurons, which is about 130 million synapses, which is still far below the 800 trillion synapses in the human brain. As members of the Intel Neuromorphic Research Com- munity (INRC), we were given access to Loihi 1 under the Kapoho Bay form factor (see Table 5), as well as the second iteration of the chip via Intel’s cloud services. Experimental results were obtained on Loihi 2.“
How can this be? Is Intel telling porkies about on chip training or are they not so much being untruthful but predicting a future iteration of Loihi 2? If the later the three year lead that AKD1000 had before AKIDA 2.0 extended it to at least five years probably needs revision.
By the way AKD1000 has on chip learning and not on chip training two very different things. On chip learning is far more advanced.
My opinion only DYOR
Fact Finder
(20min delay)