Posts

Neurons and the Nerve Net

Image
How do neurons work? Max Bennett explains in A Brief History of Intelligence . Edgar Adrian made 3 key findings on the topic in the 1920’s, for which he would win the Nobel Prize.   First , he found that neurons don’t send electric signals continuously. Rather, they fire all-or-nothing responses. On or Off, nothing in between. This raised a problematic question. Our senses can differentiate between levels of volume, strength of smells, amount of light etc. How could a simple On/Off-only mechanism convey shades?   To answer that question, Adrian took a muscle from the neck of a dead frog and attached a recording device to a single stretch-sensing neuron in that muscle. Then he did his experiment: how would the neuron convey different weights? Here is what his recording device noted:   The strength of the spike (On) did not vary with the weight. But the frequency of the spikes was proportional to the weight – higher the weight, higher the frequency of the spike...

Preamble #3: Ambedkar's Fingerprints

When the constitution was being framed, many of the members pointed out it was not assigning importance to the village as a unit of governance. Wasn’t that violating Gandhi’s view and input, they asked.   Aakash Singh Rathore’s Ambedkar’s Preamble goes into that. As mentioned in an earlier blog, Ambedkar had fallen out with Gandhi over the forced 1932 Poona Pact where he had to give up on the reservation of constituencies for the lower castes. Ambedkar was dead against the village as the smallest unit of governance because the “village is a sink of localism, a den of ignorance, narrow mindedness and communalism”.   Several members objected to this. Wasn’t this a violation of the principle of local governance, they asked. Ambedkar stood his ground pointing to Gandhi’s own admission, that “You will not understand me if you think about the villages of today… My villages… exist in my imagination”. For an uber-pragmatist like Ambedkar, governance systems could not be based...

Animal and Fungus, Neurons the Difference

A Brief History of Intelligence is written by an AI engineer who studied the brain! So why did Max Bennett study the brain? “The relations between AI and the brain goes both ways, while the brain can surely teach us about how to create artificial humanlike intelligence, AI can also teach us about the brain.” He clarifies early: it is the human brain he is referring to, not brains in general. He starts from the beginning, from the evolution of neurons .   Fungus. It is closer to animals than plants. How? Because fungus cannot do photosynthesis; it takes in food and oxygen. Yet one lineage (animals) went on to develop brains whereas the other (fungi) didn’t. Why the difference?   Well, animals and fungi adopted different strategies for food – animals kill (plants or other animals) and then digest food inside themselves. Fungi wait for things to die and then digest them outside their body. Fungi use a spray-and-pray approach – they spray trillions of singled ...

Preamble #2: Disagreements with Gandhi

The root of the “lifelong feud” between Ambedkar and Gandhi is described in Aakash Singh Rathore’s Ambedkar’s Preamble . In 1930, Ambedkar represented the “depressed classes” (the term for the lower castes) at the Round Table Conference in London and did an outstanding job. In 1932, his continuous efforts yielded results – a scheme for separate electorates for the untouchables (In present-day speak, that means constituencies reserved for the untouchables).   Gandhi resorted to a fast-onto-death against the decision, which put Ambedkar in an impossible situation. “Blackmailed into it, Dr Ambedkar signed a pact with Gandhi in 1932, with terms that were quite disagreeable to him.” It was from this point (the 1932 Poona Pact) that Ambedkar would characterize Gandhi not as a Mahatma, but as a dangerous opponent, famously describing this episode as one where Gandhi “showed me his fangs”.   Why was Gandhi so opposed to such a reservation? The book doesn’t say, but here is...

Deep Learning Overview

Neural networks. Deep learning. This is the “how” behind what we see as AI (artificial intelligence) around us today, from Alexa’s voice recognition to Google’s image search to your smartphone’s ability to unlock based on your face. While these algorithms are very powerful indeed, they are also mysterious…   “Mysterious”? What does that even mean? Simply put, we don’t understand how they conclude what they conclude. It’s not as if someone wrote specific software instructions on how to recognize a dog. Rather, a huge set of data is provided as input with tags like “dog” and “not dog”. The system goes over the data and self-discovers which patterns correspond to a dog. Remember that old joke on computers as GIGO (Garbage In, Garbage Out)? That joke was based on instructions we keyed into a computer (type the wrong instruction, you get the wrong result). With AI, GIGO now means something else – feed it the wrong data and/or wrong tags as input, and the patterns the system learn...

Preamble #1: Assorted Tidbits

The preamble to the constitution. That’s the topic of Aakash Singh Rathore’s Ambedkar’s Preamble . Not the entire constitution, just the preamble. Why a book on just the preamble? The author explains: “The ideas and principles behind a clause can be more important than the mere mechanics of the clause itself.”   It starts with something we don’t even notice this: “WE, THE PEOPLE OF INDIA… IN OUR CONSTITUENT ASSEMBLY this twenty-sixth day of November, 1949, do HEREBY ADOPT, ENACT AND GIVE TO OURSELVES THIS CONSTITUTION.” Not 26 January, 1950 (our Republic Day). Instead, 26 November 1949. And if it was ready on that earlier day, why did our Republic Day have to be later?   Therein lies a tale. In 1929, the Congress’ National Session made a call for complete independence. A while later, Gandhi published an article on 26 January 1930 saying India would settle for nothing less than “complete independence”. It was as a callback to that Gandhian demand that the coun...

DPI Design Principle #5: Privacy

Image
The last pillar of India’s DPI (Digital Public Infrastructure) is privacy . We take that to mean control over who knows what, whether that info can be shared with others, for how long it can be retained etc. Yes, privacy is all those things.   But India’s DEPA takes it a lot further, explains Rahul Matthan in The Third Way . It actively seeks to ensure portability (Unlike those private corporations which deliberately have data is in non-standard formats to prevent interoperability). Even better, DEPA has “been designed to support requests for specific items of data ”. An example helps. When we apply for say a visa, the issuing country really only wants to know if we earn enough (salary month on month). But the bank statement exposes every transaction. The DEPA framework allows you to select only salary credits be shown and everything else blacked out.   The next issue with privacy is consent , i.e., explicit permission of the individual. In theory, the solution lie...