Profile cover photo
Profile photo
Aung Thiha
Biomedical Engineer, Scientist, Transhumanist
Biomedical Engineer, Scientist, Transhumanist

Aung's posts

Post has shared content
A new kind of academic journal. This is exciting.
In case you haven't seen it, the Distill Journal (, created and edited by Google Brain team members +Christopher Olah and +Shan Carter, launched earlier this week. It's a totally new kind of journal and presentation style for machine learning research: online, interactive, and encouraging of lucid and clear presentations of research that go beyond using a Gutenberg-era presentation medium for ML topics. I'm really excited about this, and Chris and Shan have put a ton of work into this. The reactions so far have been quite amazing and positive

"Yes. Yes. Yes. A million times yes. I can't count how many times I've invested meaningful time and effort to grok the key ideas and intuition of a new AI/DL/ML paper, only to feel that those ideas and intuitions could have been explained much better, less formally, with a couple of napkin diagrams.... I LOVE what Olah, Carter et al are trying to do here." (Hacker News)

"I really love this effort. Research papers are low bandwidth way to get information into our brains..." (Hacker News)

"finally, someone gets it!! we need to COMMUNICATE research CLEARLY" (Twitter)

"My gosh, interactive dataviz is now the core of an academic journal. Thank you @shancarter & @ch402 & @distillpub!" (Twitter)

"This new machine learning journal is seriously exciting; an emphasis on clear explanation & interactive illustration" (Twitter)

"'Research Debt' - I am curious where @distillpub will go but I really like this essay by @ch402 & @shancarter" (Werner Vogels, CTO of Amazon, Twitter)

Blog posts announcing Distill:

Google Research:


Y Combinator:


Chris Olah's blog:

Post has attachment

Post has attachment
Ethical design of apps should become a big deal.
The choice of app usage is ours but thousands of the most brilliant minds on earth are working hard to break down our resistance.

Post has shared content
This is what I'd like to see next, hardware based neural networks with increasingly more complex models as time goes on, work such as what's described in the article could be used to build the first proto synthetic brains...

Post has shared content
SciTech Digest - 08/2017.
Permalink here:

MEMS AFM on-chip, Low power voice chip, Wireless power, LysoSENS development, Chiral carbon nanotubes, MOF molecular looms, Molecular biology of sleep, Electrical brain interfaces, DNA computer drugs, Printable solar cells.

1. On-chip MEMS AFM
A MEMS-based atomic force microscope has been created on a single chip complete with all of the sensors and components needed to control the device The one square centimeter sized device operates an oscillating cantilever that is moved across the surface of the sample to be imaged. While it might not have the sensitivity of a high-end laboratory system such a device should make entry-level AFM applications much cheaper and more widespread - a lot more people having access to and using an AFM can only be a good thing.

2. Low Power Voice Control Chip
A low power voice-control and speech recognition chip has been developed that achieves an energy saving of between 90% - 99%, effectively running speech-recognition software for between 0.2 - 10 milliwatts instead of the usual 1 watt that a phone uses Such low-powered capabilities are ideally suited to internet of things applications and low-power sensors and interfaces with embedded communications. The chip itself incorporates three different hardware implementations of neural networks of varying complexity.

3. Better Wireless Power Transfer
Disney research has demonstrated a quasistatic cavity resonance device for transferring power wirelessly to receivers in devices with 40% to 95% efficiency, and can transfer 1900 watts in this way safely I’ve covered several different technologies attempting to do similar wireless power transfer but this latest attempt appears to significantly improve the range, power, and efficiency. Again, a mature technology would be a key enabler of internet of things devices, sensors, and applications.

4. LysoSENS Moves Towards the Clinic
Ichor Therapeutics has demonstrated very promising results in cells for clearing types of lysosomal garbage and is now seeking to complete animal studies and move into a Phase 1 human clinical trial The therapy comes from bacterial enzymes that can break down certain types of lysosomal garbage, and which have also been modified to be targeted to the lysosomes of target cells. In this specific, niche case the therapy breaks down the garbage and removes the accumulated damage A2E metabolic waste aggregates in retinal cells that leads to different types of macular degeneration, and so represents a good, early, embryonic rejuvenation and anti-aging therapy.

5. Catalysts for Chiral Carbon Nanotubes
New work reveals that different carbon nanotube growth catalysts can preferentially form carbon nanotubes with different chiralities - the pattern of graphene hexagons around the tube that control metallic or semiconducting properties of the carbon nanotube Tungsten carbide produces semiconducting carbon nanotubes with 80% - 90% purity, while molybdenum carbide helps produce metallic carbon nanotubes. Meanwhile carbon nanotubes and graphene have been combined into functional 3D graphene rebar structures

6. Molecular Looms from MOFs
Metal Organic Framework materials are now being used to precisely position (four-armed in this case) monomer molecules that are then cross-linked in a precise array similar to two-dimensional polymer textiles This is a clever nanotechnology application for building precisely structured and formulated materials with near perfect atomic organisation. After formation the molecule-thick 2D polymer sheets are actually held together by the mechanical forces resulting from the weave pattern. A versatile platform for creating a wide variety of different, precise, 2D polymer sheets with customisable properties and structures at the atomic scale.

7. The Molecular Biology of Sleep
The molecular biology underpinning and controlling sleep is being further mapped out as part of a huge study in mice with the discovery of two new genes that play a key role in regulating sleep The first, Sik3, influences the total amount of sleep needed, while the second, Nalcn, influences the amount of REM dreaming sleep that is attained. This study took years and involved mutating the genes of thousands of mice and hooking them up to brainwave monitors while they slept. With these targets identified there is further scope to rationally design interventions able to modify sleep in humans.

8. Better Electrical Brain Interfaces
In just one week we had three different improved electronic brain interfaces announced. First, a new complementary metal oxide semiconductor nanoelectrode array can image and map the changing electrical signals within a large group of living cells Second, ultra-flexible nanoelectronic threads can act as reliable brain probes that enable scar-free integration for neural signal recording Finally, glassy-carbon electrodes transmit more robust signals to restore function in people with damaged spinal cords

9. DNA Computer Smart Drugs
A new DNA computer is able to process the presence and concentration of multiple specific antibodies in the body at once in order to diagnose particular disease states or see the paper In this process DNA strands are designed to bind to different antibodies, and when mixed with complementary reporter DNA sequences, these sequences only release the “signal” strand when those specific antibodies are present. These output signals are then processed by a range of DNA computer and logic elements to provide information on the nature of the disease that is present. I’m impressed by how sophisticated the DNA computing and health diagnostics platform is becoming.

10. Printable Perovskite Solar Cells
A new chemical reaction allows an electron-selective solar cell layer to be grown in solution out of nanoparticles directly on top of electrodes and that also incorporate perovskite solar-power ink, and at much lower temperatures than was previously possible. The solar cells created with this process in the lab demonstrated an energy efficiency of 20.1%. The promise of printable solar cells is being able to cheaply produce high-efficiency panels via established printing techniques or even custom-printing onto most desired surfaces.

SciTech Tip Jar:
Animated Photo
2 Photos - View album

Post has attachment

Post has attachment
Microbial mind control?

Post has shared content

Post has shared content

Post has shared content
SciTech Digest - 04/2017.
Permalink here:

Atmospheric laser lenses, EMP artillery, Virus communications, Smartphone DNA sensing, Learning software creates learning software, Functionally imprinted polymers, Better RNA aptamers, Manipulating the vacuum, Safer Tesla autopilot, Filtering radioactivity.

1. Atmospheric Laser Lenses
The atmosphere can play havoc with light and lasers but a Laser Developed Atmospheric Lens system proposes methods to use powerful lasers to turn regions of the atmosphere into lenses, mirrors, and “deflector shields” This system would produce ordered hot and cold layers or structures of air in order to control the refractive conditions that light must pass through. This might be used to produce large atmospheric lenses for terrestrial or space-based telescopes, or a region that disperses incoming lasers and directed energy and act as a shield. This would spark an arms race of defensive and offensive capabilities.

2. EMP Artillery
In related news artillery shells capable of delivering a targeted EMP or ElectroMagnetic Pulse are being developed These munitions are intended of course to knock out all electronics within a region. While shielding may protect certain key electronics, those driving any wireless communications when the EMP hit would of course be knocked out. Of course, the atmospheric lenses from above may also provide protection from such EMP blasts (depending on strength) as well as modifying the effects of an EMP in an offensive manner.

3. Virus Communication System Discovered
A viral communication system has been discovered for the first time in which viruses sense chemical signals left behind by earlier viral copies to decide whether to kill or just infect their bacterial hosts Evolutionarily this makes sense, because if the virus is running out of bacterial hosts (having killed too many), then it is best to insert into the host cell’s genome and await reactivation at a later date to re-establish growth. This mechanism might also be present in viruses that infect human cells that have an active and dormant phase, such as HIV and HSV; forcing the viruses to stay dormant via a drug would be therapeutically beneficial.

4. DNA Mutation Detection and Molecular Diagnostics via Smartphone
A new 3D printed device combines sample wells, a moveable stage, lenses, and laser diodes, and docks with your mobile phone to allow your phone camera to detect mutations present in the DNA of cells in the sample as part of performing remote, distributed, point of care molecular diagnostic analysis Cells of interest are loaded into the device along with fluorescent DNA probes for mutations and DNA sequences of interest, and after following the protocol the phone camera can confirm via imaging the sample whether certain sequences are present or not. These might be as common as thermometers at home one day.

5. Machine Learning System Creates New Machine Learning Systems
Researchers at Google Brain have developed machine learning system that designed another machine learning system that was able to run a language-processing software benchmark better than software designed by humans So here we have learning software making learning software, which can be expected to more rapidly disseminate machine learning tools and capabilities across the digital ecosystem. Other experiments demonstrated learning systems creating new learning systems with a better ability to generalise and able to master new tasks with less additional training than usual. Meanwhile machine learning is getting ever-better at creating pop music

6. Functionally Imprinted Polymers
Novel polymers can now be functionally imprinted with DNA molecules, and retain the ability to bind to that specific DNA sequence In this process the target (DNA or other molecule) is added to a solution of special monomers that assemble around the target before being electrochemically polymerised. Applications for such molecular imprinting include creating recognition films for chemical sensors and also in for purification of solutions, removing specific contaminants and other molecules as needed.

7. Evolving Better RNA Aptamers
A better way to evolve more effective and stable RNA aptamers has been discovered simply by utilising naturally occuring stable RNA structures as a starting point As part of the proof-of-concept the group used pieces of natural riboswitches and ribozymes as scaffolds to evolve RNA aptamers that bind target molecules of interest such as amino acids and other small molecules. This work provides a robust set of design principles that others can use to quickly generate RNA aptamers that specifically bind target molecules of interest, and which fold correctly and remain stable in cells, which was a major problem solved by this work.

8. Manipulating the Vacuum Group State
As part of a new way to study the quantum vacuum ground state physicists claim to demonstrate the detection of signals from completely empty space In this work femtosecond light pulses are used to probe electromagnetic fluctuations that apparently lack intensity, probing discrete time points instead of discrete frequency bands, and these are believed to be vacuum fluctuations. The thrust of the work appears to be the direct detection of electromagnetic background noise & fluctuations in a small region of space. They claim to be able to manipulate the vacuum in this way such that the measured noise is lower than the conventional vacuum ground state. The group aim to determine whether this qualifies as a weak measurement system that leaves quantum systems unperturbed. I wonder if such a method might help with “hidden variables” type experiments.

9. Tesla’s Autopilot Ten Times Safer
Federal data analysis reveals that Tesla’s first generation Autopilot reduced the crash rate incidence by 40%, while the second generation software is set to reduce the crash rate by 90%, or ten times lower than human drivers This software is built on 1.3 billion miles of data collected from the suite of sensors carried by the cars and the 90% reduction rate is made possible by new upgradeable hardware introduced in October 2016. It only has to be better than human drivers at saving lives; ten times better (and beyond that in future) is so compelling it cannot be ignored.

10. Extracting Radioactive Elements from Water
An inexpensive “oxidatively modified carbon” material has been developed that is efficient at absorbing radioactive elements including cesium and strontium from contaminated water Conventional absorbents often have to be stored as nuclear waste, but this material can be burnt in a nuclear furnace to produce a much smaller amount of radioactive ash for storage and handling. Cleaning contaminated water and soil would be a great application of course, but I also wonder about the utility here for extraction and harvesting of particular elements from the oceans for example.

Bonus: I couldn’t pass up this fascinating method for making huge soap bubbles and the accompanying discussion on using modifications to this for manufacturing huge complex structures in space

SciTech Tip Jar:

Animated Photo
2 Photos - View album
Wait while more posts are being loaded