Suramya's Blog : Welcome to my crazy life…

August 26, 2022

Using MultiNerf for AI based Image noise reduction

Filed under: Computer Software,Emerging Tech,My Thoughts,Tech Related — Suramya @ 2:58 PM

Proponents of AI constantly come up with claims that frequently don’t hold up to extensive testing, however the new release from Google Research called MultiNerf which runs on RAW image data to generate what the photos would have looked like without the video noise generated by imaging sensors seems to be the exception. Looking at the video it almost looks like magic, and appears to work great. Best of all, the code is open source and already released on GIT Hub under the Apache License. The repository contains the code release for three CVPR 2022 papers: Mip-NeRF 360, Ref-NeRF, and RawNeRF.

TechCrunch has a great writeup on the process. DIYPhotography has created a video demo of the process (embedded below) that showcases the process:


Video Credits: DIYPhotography

I like the new tools to make the photographs come out better, but I still prefer to take unaltered photos whenever I can. The most alteration/post-processing that I do on the photos is cropping and resizing. That also is something I do infrequently. But this would be of great use to professional photographers in conditions that are less than optimal.

– Suramya

August 6, 2022

Post Quantum Encryption: Another candidate algorithm (SIKE) bites the dust

Filed under: Computer Security,Computer Software,Quantum Computing — Suramya @ 8:23 PM

Quantum Computing has the potential to make the current encryption algorithms obsolete once it gets around to actually being implemented on a large scale. But the Cryptographic experts in charge of such things have been working on Post Quantum Cryptography/Post Quantum Encryption (PQE) over the past few years to offset this risk. SIKE was one of KEM algorithms that advanced to the fourth round earlier this year and it was considered as an attractive candidate for standardization because of its small key and ciphertext sizes.

Unfortunately while that is true researchers have found that the algorithm is badly broken. Researchers from the Computer Security and Industrial Cryptography group at KU Leuven published a paper over the weekend “An Efficient Key Recovery Attack on SIDH” (Preliminary Version) that describes a technique which allows an attacker to recover the encryption keys protecting the SIKE Protected transactions in under an hours time using a single traditional PC. Since the whole idea behind PQE was to identify algorithms that are stronger than the traditional ones this immediately disqualifies SIKE from further consideration.

Abstract. We present an efficient key recovery attack on the Supersingular Isogeny Diffie–Hellman protocol (SIDH), based on a “glue-and-split” theorem due to Kani. Our attack exploits the existence of a small non-scalar endomorphism on the starting curve, and it also relies on the auxiliary torsion point information that Alice and Bob share during the protocol. Our Magma implementation breaks the instantiation SIKEp434, which aims at security level 1 of the Post-Quantum Cryptography standardization process currently ran by NIST, in about one hour on a single core.

The attack exploits the fact that SIDH has auxiliary points and that the degree of the secret isogeny is known. The auxiliary points in SIDH have always been an annoyance and a potential weakness, and they have been exploited for fault attacks, the GPST adaptive attack, torsion point attacks, etc.

This is not a bad thing as the whole testing and validation process is supposed to weed out weak algorithms and it is better to have them identified and removed now than after their release as then it becomes almost impossible to phase out systems that use the broken/compromised encryption algorithms.

Source: Schneier on Security: SIKE Broken

– Suramya

May 27, 2022

Creating robots with no moving parts or computational ability which can navigate through mazes on their own

Filed under: Emerging Tech,Science Related — Suramya @ 11:34 PM

One would imagine that it takes skill or at least the ability to think to escape from a maze, unless you count running around like a headless chicken as a skill. However, Jie Yin and his colleagues at North Carolina State University have created a contraption that has no computational ability or moving parts but is still able to escape from a maze using trial and error.

The device is shaped like a pasta and is made from a rubber like material impregnated with liquid crystals. When this device is placed on a heated surface the parts in contact with the surface heat up and expand while the rest of the device remains the same this causes a twisting motion that allows it to roll at a speed of up to 3.8 millimetres per second. Even more interestingly this ‘robot’ can navigate a maze, when it reaches an obstacle such as a wall its orientation changes slightly and can sometimes continue moving. If that doesn’t work, then it continues to push against the obstacle which creates changes in the tension in the device allowing it to change the orientation of the arc’s on its surface to another direction, which would enable it to roll in the opposite direction. These two abilities enable it to continually change direction when meeting obstacles, bumping from surface to surface, eventually finding its way out despite lacking any intelligent control.

Autonomy is crucial for soft robotics that are constructed of soft materials. It remains challenging to create autonomous soft robots that can intelligently interact with and adapt to changing environments without external controls. To do so, it often requires an analogical soft “brain” that integrates on-board sensing, control, computation, and decision-making. Here, we report an autonomous soft robot embodied with physical intelligence for decision-making via adaptive soft body-environment interactions and snap-through instability, without integrated sensing and external controls. This study harnesses physical intelligence as a new paradigm for designing autonomous soft robots that can interact intelligently with their environments, thus potentially reducing the burdens on the conventional integrated sensing, control, computations, and decision-making systems in designing intelligent soft robots.

Soft robots that can harvest energy from environmental resources for autonomous locomotion is highly desired; however, few are capable of adaptive navigation without human interventions. Here, we report twisting soft robots with embodied physical intelligence for adaptive, intelligent autonomous locomotion in various unstructured environments, without on-board or external controls and human interventions. The soft robots are constructed of twisted thermal-responsive liquid crystal elastomer ribbons with a straight centerline. They can harvest thermal energy from environments to roll on outdoor hard surfaces and challenging granular substrates without slip, including ascending loose sandy slopes, crossing sand ripples, escaping from burying sand, and crossing rocks with additional camouflaging features. The twisting body provides anchoring functionality by burrowing into loose sand. When encountering obstacles, they can either self-turn or self-snap for obstacle negotiation and avoidance. Theoretical models and finite element simulation reveal that such physical intelligence is achieved by spontaneously snapping-through its soft body upon active and adaptive soft body-obstacle interactions. Utilizing this strategy, they can intelligently escape from confined spaces and maze-like obstacle courses without any human intervention. This work presents a de novo design of embodied physical intelligence by harnessing the twisting geometry and snap-through instability for adaptive soft robot-environment interactions.

This technology could be used to create cheap robots that can explore environments to take sensor readings and can potentially function inside the human body when made in microscopic scale. Since they don’t have any moving parts and don’t require power sources it would allow them to function for a longer duration than powered alternatives which would eventually run out of power. Plus, since they don’t require batteries it would be safer for people to ingest them without potentially harmful effects because most of the power sources in use today have some harmful chemicals in them.

The team’s findings have been published in the Proceedings of the National Academy of Sciences (PNAS) Journal: Twisting for soft intelligent autonomous robot in unstructured environments earlier this week.

Source: New Scientist: Pasta-shaped robot with no moving parts can navigate through mazes

– Suramya

May 14, 2022

Using algae sealed in a AA battery to generate enough electricity to run a microprocessor for 6 months

Powering computers and all our devices requires us to use batteries if they can’t be connected to a power source/electrical socket. For the most part this means that we use NiCa or Lithium batteries. The problem with this is that they require us to use rare earth metals that are hard to find and process, which makes them expensive and mining the metals are potentially bad for the environment. The other problem is that they need frequent replacement and create a lot of waste. Due to this a lot of effort is going on to find better ways of generating power.

Now, Christopher Howe and other researchers from the University of Cambridge have managed create a power source using blue-green algae to generate enough electricity to power a processor performing calculations (to simulate load). Using a type of cyanobacteria called Synechocystis sp. PCC 6803 sealed in a container about the size of an AA battery, made of aluminum and clear plastic they were able to generate the 0.3 microwatts of power to run the CPU for 45 minutes followed by 15 minutes of standby, which required 0.24 microwatts of power.

The system ran without additional intervention for 6 months and the computer was placed on a windowsill at one of the researchers’ houses during the test and the ambient light was enough to power the processor. There are indications that this can be scaled up to generate more power for more resource intensive applications but even if that doesn’t work out, the current setup could potentially be used to power IoT devices that don’t require that much power to run such as sensors/monitors deployed in the forests/cities for monitoring.

Sustainable, affordable and decentralised sources of electrical energy are required to power the network of electronic devices known as the Internet of Things. Power consumption for a single Internet of Things device is modest, ranging from μW to mW, but the number of Internet of Things devices has already reached many billions and is expected to grow to one trillion by 2035, requiring a vast number of portable energy sources (e.g., a battery or an energy harvester). Batteries rely largely on expensive and unsustainable materials (e.g., rare earth elements) and their charge eventually runs out. Existing energy harvesters (e.g., solar, temperature, vibration) are longer lasting but may have adverse effects on the environment (e.g., hazardous materials are used in the production of photovoltaics). Here, we describe a bio-photovoltaic energy harvester system using photosynthetic microorganisms on an aluminium anode that can power an Arm Cortex M0+, a microprocessor widely used in Internet of Things applications. The proposed energy harvester has operated the Arm Cortex M0+ for over six months in a domestic environment under ambient light. It is comparable in size to an AA battery, and is built using common, durable, inexpensive and largely recyclable materials.

Their research has been published in the Energy & Environmental Science journal and work is ongoing to build on top of it to look at commercial applications.

Source: A colony of blue-green algae can power a computer for six months

– Suramya

May 10, 2022

Using ancient techniques for adding secret images in bronze mirrors to hide images in Liquid Crystal displays

Filed under: Emerging Tech,Interesting Sites,Science Related — Suramya @ 1:28 AM

There are a lot of things that were accomplished by our ancestors that seem like they should be impossible and this is why the theory that aliens were involved in our past to give us a boost is so popular. People don’t realize that just because it wasn’t possible in the western world doesn’t mean that others in the world couldn’t do it. In this post I am going to talk about Chinese/Japanese Magic mirrors that were first created ~200BC but modern science was only able to explain how they work in 2005 when M V Berry published an paper describing the optics of how this would work.

The Magic Mirror is a type of mirror that was popular in ancient china, specially the Han dynasty (206 BC – 24 AD). The specialty of these mirrors is that they were made out of solid bronze with the front side polished brightly so that it can be used as a mirror whereas the back would have a design cast in the metal. When a bright light was reflected by the mirror and shone against a wall the pattern on the back of the mirror would be projected onto the wall.


Example of how the Magic Mirror reflections look (Pic credit: Faena.com)

As you can imagine this is extremely hard to do. Due to trading with the Chinese, folks over in Korea and Japan have also been known to create these mirrors which are known as Makyō (magic mirrors) over there. One difference between Makyō and the Chinese mirror is that a Makyō doesn’t reflect the image on the back on the mirror when light hits it, nor does it have any obvious irregularities on its reflecting surface. But still it creates these fantastical images where nothing should be there. More details on how the mirrors were constructed and the history behind them are available here.

It took western scientists over 2000 years to figure out the science behind these mirrors, kind of.. as evident from the explanation below.

Although the surface of the mirrors is polished and seems completely flat, it has subtle convex and concave curves caused by the designed. Convex curves (outwards) scatter light and darken their areas of reflection. For their part, concave curves focus light and illuminate their areas of reflection. Mirrors are made of forged bronze, and the thickest parts are cooled at a different speed than the thin ones. Since the metal contracts a little as it is cooled, the different ranges of cooling “stress” or slightly deform the metal. The thin areas are also more flexible than the thick parts, so the polishing process, which should smoothen the metal until uniformity is achieved, exaggerates the slight differences in thickness. While we cannot see the pattern on the surface of the mirror, photos very clearly delineate it, so when they are able to bounce off the mirror’s curves, the pattern emerges.

Using the understanding gained from Berry’s paper Felix Hufnagel and his colleagues from the University of Ottawa in Canada to create a modern version of the magic mirror using liquid crystal which is a different state of matter (their molecules are both fluid and arranged in patterns). By applying an electric current to the liquid crystals they were able to tailor the orientation of the molecules which allowed them to create an image which would only show up when a particular combination of current/amplitude was applied. The images created using this technique look clear even when viewed from different angles which can be used to improve projectors for 3D images.

Their paper was published in Optica earlier this month and is an interesting (if confusing read).

Interesting links:
Wikipedia: Chinese Magic Mirror
Secret images hidden in mirrors and windows using liquid crystals

– Suramya

May 9, 2022

Researchers have created the first one-way superconductor which could lower energy used by computers

Filed under: Computer Hardware,Emerging Tech,Science Related,Tech Related — Suramya @ 6:58 PM

Computers use massive amounts of energy worldwide and with the increasing dependence on computers in our life the energy utilization is only going to go up. To give you an idea, the International Energy Agency estimates that 1% of all global electricity is used by data centers. There are multiple efforts ongoing to reduce the power consumption and the recent advances by Mazhar Ali from Delft University of Technology in the Netherlands and his colleagues are a great step forward in this direction.

Mazhar and team have successfully demonstrated a working superconducting diode by sandwiching a 2D layer of a material called niobium-3 bromine-8, which is thought to have a built-in electric field, between two 2D superconducting layers. When electrons travel through the structure in one direction, they don’t encounter resistance, but in the other direction they do. This is unique because till now we had only gotten a diode working with non-superconducting metals (as they would not give any resistance in either direction).

The superconducting analogue to the semiconducting diode, the Josephson diode, has long been sought with multiple avenues to realization being proposed by theorists1,2,3. Showing magnetic-field-free, single-directional superconductivity with Josephson coupling, it would serve as the building block for next-generation superconducting circuit technology. Here we realized the Josephson diode by fabricating an inversion symmetry breaking van der Waals heterostructure of NbSe2/Nb3Br8/NbSe2. We demonstrate that even without a magnetic field, the junction can be superconducting with a positive current while being resistive with a negative current. The ΔIc behaviour (the difference between positive and negative critical currents) with magnetic field is symmetric and Josephson coupling is proved through the Fraunhofer pattern. Also, stable half-wave rectification of a square-wave excitation was achieved with a very low switching current density, high rectification ratio and high robustness. This non-reciprocal behaviour strongly violates the known Josephson relations and opens the door to discover new mechanisms and physical phenomena through integration of quantum materials with Josephson junctions, and provides new avenues for superconducting quantum devices.

The next step is to create a superconducting transistor, but there are multiple challenges ahead that need to be overcome before this can be commercially released. The first problem is that the diode only works when it’s temperature is at 2 kelvin, or -271°C which uses more energy than the diode saves. So the team is looking at alternative materials so that they can get it to work at 77 Kelvin (which is when nitrogen is liquid) so the energy used would be less and we would have an energy-saving diode.

Another issue to be sorted is that the current process of making the diode is manual and would need to be automated for large scale production. But that is a future problem as they first need to find a combination of materials that works at a reasonable energy cost.

Source: First one-way superconductor could slash energy used by computers
Paper: The field-free Josephson diode in a van der Waals heterostructure

– Suramya

May 2, 2022

MIT researchers create a portable desalination unit that can run off a single solar panel

Filed under: Emerging Tech,My Thoughts,Science Related — Suramya @ 2:33 AM

The lack of drinking water is a major problem across large portions of the world and over 2 billion people live in water-stressed countries. According to WHO at least 2 billion people use a drinking water source contaminated with feces. On the other side, places near the sea have to deal with salt water contamination of their drinking supply. If we can desalinize sea water cheaply and easily then it will be a great boon to world.

There are existing technologies that convert sea-water to drinking water but they require massive energy supply and large scale plants which are very expensive to make. To resolve this issue MIT researchers have been working on creating a portable desalination unit that generates clear, clean drinking water without the need for filters or high-pressure pumps. Since the unit doesn’t use filters or high-pressure pumps the energy requirement is low enough that it can be run off a small, portable solar panel.

The research team of Jongyoon Han, Junghyo Yoon, a research scientist in RLE; Hyukjin J. Kwon, a former postdoc; SungKu Kang, a postdoc at Northeastern University; and Eric Brack of the U.S. Army Combat Capabilities Development Command (DEVCOM) created this and the initial prototype has worked as expected. Their research has been published online in Environmental Science and Technology.

Instead, their unit relies on a technique called ion concentration polarization (ICP), which was pioneered by Han’s group more than 10 years ago. Rather than filtering water, the ICP process applies an electrical field to membranes placed above and below a channel of water. The membranes repel positively or negatively charged particles — including salt molecules, bacteria, and viruses — as they flow past. The charged particles are funneled into a second stream of water that is eventually discharged.

The process removes both dissolved and suspended solids, allowing clean water to pass through the channel. Since it only requires a low-pressure pump, ICP uses less energy than other techniques.

But ICP does not always remove all the salts floating in the middle of the channel. So the researchers incorporated a second process, known as electrodialysis, to remove remaining salt ions.

Yoon and Kang used machine learning to find the ideal combination of ICP and electrodialysis modules. The optimal setup includes a two-stage ICP process, with water flowing through six modules in the first stage then through three in the second stage, followed by a single electrodialysis process. This minimized energy usage while ensuring the process remains self-cleaning.


Video demonstration of the process

The prototype device was tested at Boston’s Carson Beach and was found to generate drinking water at a rate of 0.3 liters per hour, requiring only 20 watts of power per liter during the use. As you can guess this is pretty amazing. If the device can be mass-produced it will help reduce the scarcity of drinking water in the world without requiring massive amounts of energy which would cause other climate impact.

One downside of this kind of machine is that it creates a byproduct of highly saline water as the salt from the pure water is mixed with the waste water. Releasing this water in the ocean has a huge impact on the sea life as the water suddenly becomes too saline for them. If the water is allowed to seep into the land then it will reduce the fertility of the soil due to the increased salt in the soil. In addition to making the device commercial we also need to do research on what we should do with the waste water generated so that the adverse impact of the product can be offset.

Source: MIT News: From seawater to drinking water, with the push of a button

– Suramya

April 27, 2022

MIT’s Ultra-thin speakers can be used to make any surface into a low-power, high-quality audio source

Filed under: Computer Hardware,Emerging Tech,Tech Related — Suramya @ 9:51 PM

Noise Cancellation is one of those things that initially we think that we don’t need but once you start using it, it becomes indispensable. I got my first set of noise canceling headsets back in 2002-2003 when I had a coworker who was extremely loud and would insist on sharing their thoughts in a very loud voice. The cherry on top was that a lot of what they said was wrong and it would grab my attention. I would be peacefully working then I would hear something and be like did they just make this statement? In short it was very distracting. So I got a noise canceling headset and was able to ignore them. Since then I have ensured that I always have my noise canceling headsets handy both at work and while traveling.

But you can’t install noise canceling everywhere (at least not cheaply). I have been fortunate that most of the places I have stayed at I didn’t have the problem of loud neighbors but others are not as fortunate. Loud neighbors are one of the major problems in urban life. Which is why I love this new invention by the folks over at MIT that allows you to convert your entire wall into a noise cancelling surface by putting ultra-thin speakers as a wallpaper in your room. These speakers are very thin & use very little power (100 milliwatts of electricity to power a single square meter).

their design relies on tiny domes on a thin layer of piezoelectric material which each vibrate individually. These domes, each only a few hair-widths across, are surrounded by spacer layers on the top and bottom of the film that protect them from the mounting surface while still enabling them to vibrate freely. The same spacer layers protect the domes from abrasion and impact during day-to-day handling, enhancing the loudspeaker’s durability.

To build the loudspeaker, the researchers used a laser to cut tiny holes into a thin sheet of PET, which is a type of lightweight plastic. They laminated the underside of that perforated PET layer with a very thin film (as thin as 8 microns) of piezoelectric material, called PVDF. Then they applied vacuum above the bonded sheets and a heat source, at 80 degrees Celsius, underneath them.

Because the PVDF layer is so thin, the pressure difference created by the vacuum and heat source caused it to bulge. The PVDF can’t force its way through the PET layer, so tiny domes protrude in areas where they aren’t blocked by PET. These protrusions self-align with the holes in the PET layer. The researchers then laminate the other side of the PVDF with another PET layer to act as a spacer between the domes and the bonding surface.

The applications are endless for this technology. They can be used to soundproof apartments, planes, cars etc. They can be used to create 3D immersive experiences cheaply without having to install gigantic speakers. They could also be used in phones and other devices to play sound/music. Since they are paper-thin, we can apply them as a wallpaper in a room that can be removed when moving out, which would allow renters to install them in the apartments.

The work is still in its early stages but it looks very promising.

Source: Gizmodo: Cover Your Wall in MIT’s New Paper Thin Speakers to Turn Your Bedroom Into a Noise Cancelling Oasis

– Suramya

April 25, 2022

Rainbow Algorithm (one of the candidates for post-quantum Cryptography) can be broken in under 53 hours

Quantum Computing has the potential to make the current encryption algorithms obsolete once it gets around to actually being implemented on a large scale. But the Cryptographic experts in charge of such things have been working on Post Quantum Cryptography over the past few years to offset this risk. After three rounds they had narrowed down the public-key encryption and key-establishment algorithms to Classic McEliece, CRYSTALS-KYBER, NTRU, and SABER and te finalists for digital signatures are CRYSTALS-DILITHIUM, FALCON, and Rainbow.

Unfortunately for the Rainbow algorithm, Ward Beullens at IBM Research Zurich in Switzerland managed to find the corresponding secret key for a given Rainbow public key in 53 hours using a standard laptop. This would allow anyone with a laptop to ‘prove’ they were someone else by producing the secret key for a given public key.

The Rainbow signature scheme [8], proposed by Ding and Schmidt in 2005, is one of the oldest and most studied signature schemes in multivariate cryptography. Rainbow is based on the (unbalanced) Oil and Vinegar signature scheme [16, 11], which, for properly chosen parameters, has withstood all cryptanalysis since 1999. In the last decade, there has been a renewed interest in multivariate cryptography, because it is believed to resist attacks from quantum adversaries. The goal of this paper is to improve the cryptanalysis of Rainbow, which is an important objective because Rainbow is currently one of three finalist signature
schemes in the NIST Post-Quantum Cryptography standardization project.

This obviously disqualifies the algorithm from being standardised as it has a known easily exploitable weakness. It goes on to prove that cryptography is not easy and the only way to ‘prove’ the strength of an algorithm is to let others test them for vulnerabilities. Or as Bruce Schneier put it in Schneier’s Law: ‘Anyone can create an algorithm that they themselves can’t break.’ , you need others to validate that claim.

Paper: Breaking Rainbow Takes a Weekend on a Laptop by Ward Beullens (PDF)
Source: New Scientist: Encryption meant to protect against quantum hackers is easily cracked

– Suramya

April 24, 2022

Smart-contract bug locks away $34 million highlighting major weakness in smart-contracts

Filed under: Computer Software,Emerging Tech,Tech Related — Suramya @ 9:57 PM

Over the years I have had many conversations with people about Blockchain and how it is supposed to solve all our problems, but for the most part I think Blockchain is overrated and doesn’t solve any problem that can’t be solved in an easier way using less resources. Then as if Blockchain’s were not enough someone went and created smart contracts which are basically programs stored on a blockchain that run when predetermined conditions are met. They typically are used to automate the execution of an agreement so that all participants can be immediately certain of the outcome, without any intermediary’s involvement or time loss. They can also automate a workflow, triggering the next action when conditions are met.IBM Smart-Contracts Def

The major issue with a blockchain contract is that the contract is immutable so if there is a bug in the program no one can modify it to fix the issue. When warned about this potential problem the proponents of the smart-contract pretty much handwaved the concerns away stating that the issue is not that big a deal and people were just opposing them because they dislike smart-contracts and are sticks in mud etc etc.

Unfortunately, this is no longer a theoretical issue as the developers of the AkuDreams contract found out over the weekend. Due to a bug in the contract code $34 million, or 11,539 eth, is permanently locked into the AkuDreams contract forever. It cannot be retrieved by individual users or by the dev team..

This shows how dangerous it is to have a program that can’t be modified because no matter what we do we can’t ensure that code written will be 100% bug free in all the cases. When there is a bug in regular software be can push out a patch to fix it, but that is not an option for smart-contracts and that as you can see becomes an expensive issue.

Source: $34M permanently locked into AkuDreams contract forever due to bad code

– Suramya

« Newer PostsOlder Posts »

Powered by WordPress