The post IMG extends its relationship with Forbidden Technologies in 12 month deal appeared first on Blackbird.

The post Beyond Live Streaming – brand benefits & monetisation appeared first on Blackbird.

The post Filling the void before Video over IP appeared first on Blackbird.

The post Meltdown and Spectre appeared first on Blackbird.

The post Corrivium applies Blackbird video technology to roster of VOD delivery appeared first on Blackbird.

I have made a priority of engaging with our existing customers across North America and the UK. These range from MSG Networks and Viacom in New York, to Envy and the post production houses in Soho, to Deltatre, IMG, Gfinity, and the BBC. Other major clients and household name sports leagues and brands are covered by
Non-Disclosure Agreements.

Being from the broadcasting world, I already had a good understanding of the current issues impacting the industry. However, these customer visits have highlighted the challenges and opportunities in the industry as it undergoes seismic digital transformation. We are talking to existing and potential customers about how Forbidden Technologies can help increase speed to market, reduce hardware footprint, provide better video content visibility and contribute to media distribution in an evolving digital landscape. We have also held discussions with many prospective customers on both sides of the Atlantic, where I have gained valuable insight into the power of Blackbird.

Blackbird is our visionary piece of patented video technology that differentiates Forscene. Invented and developed by Stephen Streater, Blackbird is a high-performance video codec. It provides a responsive viewing experience and frame accurate manipulation of content from multiple locations across the media supply chain. Blackbird runs effectively in the cloud, on premise and in hybrid solutions. It is incredibly impressive and is what sets us apart. By selling our technology purely as an editor you could say we have been hiding this ‘light under a bushel’!

Surrounding Blackbird is a suite of microservices that take advantage of this unique technology. Our video platform, Forscene, helps solve many of the challenges faced and provides new opportunities for companies in relation to their live, non-live and archived content. Forscene is interoperable and can be integrated into existing and developing media supply chains.

Feedback from both existing and prospective customers is unanimous – Blackbird video technology should be a game changer within the industry – reaffirming my core reason for joining Forbidden as CEO and investing personally in the company. Throughout my media career, and especially in my time at Turner and BBC Worldwide, I developed a deep appreciation of the need to continuously improve speed and efficiency, which Blackbird is ideally positioned to do.

Our primary commercial objective is to integrate Blackbird as an infrastructure component in the media supply chain. These infrastructure sales have far more financial potential, although they have longer sale cycles. There is a lot of work to be done, to tackle both the lack of sales and the fact that Forbidden has been under-selling its technology, but I relish a challenge, and we are making good headway for the new year ahead.

Although we have replaced some of the sales and marketing team, we are still actively recruiting for additional sales personnel for both the UK and North America, to align with Forbidden’s new direction. The new team will report to the Sales Director who joined the company in July 2017. A new senior Marketing Manager has already been engaged. Forbidden is also looking to extend its reseller efforts to the West Coast USA and LatAm.

Lastly and most importantly, on a commercial note, over the past few months, we have signed a number of exciting new strategic Proofs of Concept (PoCs) for Blackbird in multiple regions and new sectors. This adds up to easily the most POCs the company has ever had at any one time.

These are exciting times at Forbidden Technologies and we look forward to keeping you up-to-date in the new year.

Have a Merry Christmas and Happy New Year 2018
Ian McDonough and the Forbidden team

The post Looking Forward appeared first on Blackbird.

In the old days, people would spend a long time teaching computers rules. Expert systems attempted to simulate the expertise of professionals by following the same rules. This has several drawbacks:

• experts don’t always understand the rules they use;
• many rules experts use are wrong; and
• takes a long time to teach a computer rules.

Rather than remembering facts or fixed responses to questions, an AI needs a model of its world. It’s not enough to know some good chess moves – an AI needs to know what moves are possible and to be able to assess how good they are. Once it has a model, it can make its own predictions, explore possible outcomes, and build new refinements automatically.
This model is informed – and preferably inferred – from training data. When AI was getting started, the internet was packed with text – tainted by human bias. The easiest way to get raw training data was to use live video – a rare type of readily available fresh information to analyse and understand – so this was the logical place for me to start.

Eidos

Plato conjectured that the imperfect physical world was mirrored by a world of forms – εἶδος (eidos). In summary, these forms are the pure essence of the physical realisation of the real world.
My AI model comprises forms, and the relationships between them.
Being made of pure information, these forms are amenable to processing by computer. The relationships between these forms can be studied in abstraction – and these relationships correspond to relationships in the real world.
The first step is to convert the real world into forms which can be uploaded into a computer. What do these forms look like? Being on a computer, they look like 0s and 1s. Or rather, whether each object has the property of this form is a 0 or a 1.
I split the world into “Low-level concepts” and “High-level concepts”. Low-level concepts are characteristics which the computer can easily check an object has, each using a program which returns a 0 or a 1. Examples are whether a region is light or dark, big or small, smooth or textured, inside another region. “High-level concepts” are human recognisable concepts, like a bridge, a won chess position or a catchy song.
The AI can also make up its own concepts – “Mid-level concepts”. Many concepts are related, and you can simplify combinations into single concepts. “Wheel-like” may contain many features which, in various combinations, makes it more likely you have a wheel eg round, has a rubber tyre, has spokes, has metal in it. This packaging of multiple simple concepts into higher level concepts is fundamental to intelligence.
So the question for the computer is: what is the relationship between the things you can measure (low-level concepts), and the concepts you want the computer to understand (high-level concepts)?
Luckily, you can enumerate every possible relationship. This is an exponential job – bigger than that even – but you can list these ideas in order of complexity. This allows you to start at the beginning, finding the easy ideas, and continue to more and more complex ideas. This process is a one-off investment – and easily parallelisable.
The connections between real world data are enumerated and analysed. And the more the training data, the more the truth shines out above the noise in the real world data. It just takes time.
And not as long as you might think. Amongst all the billions of ideas is the best one – the one nearest the perfect form. And this works pretty well. All you have to do is to find it.

Statistics

Mathematics is an incredibly powerful tool, and Statistics comes in handy here. Statistical significance can tell you which ideas represent a real connection, and which are just a chance combination of inputs.

Information theory

I love Information Theory. I was lucky enough to meet and discuss it with its brilliant inventor, Claude Shannon, at a computer games competition.
While Statistics tells the AI which of the ideas it conceives are more than just a chance coincidence, Information Theory tells it how much information each idea contains.
The results may shock a casual reader, used to terabyte (1,000,000,000,000 bytes) disks, with each byte equivalent to 8 bits, and each bit enough information to answer the question “did the coin toss come out heads or tails”. Surprisingly, the question “did the die come out a 6” requires, on average, less than 1 bit to answer – about 0.65 bits. This is because you already know the answer is probably “no”, so it takes less than 1 bit to confirm this (on average). If you are looking for a rare event, even the most accurate test can only give you a fraction of a bit. Did my ticket win the national lottery? About a millionth of a bit will tell you!
The AI discards billions of inferior ideas, and keeps just the best. Surprisingly (or maybe unsurprisingly, given the process) these are good enough to recognise complex real world concepts from simple inputs.

Unlimited intelligence?

Can we look at more and more complex ideas until we have an unlimited intelligence? Yes – and no.
The more ideas we look at, the more subtle the connections we discover. But if we go too far, the limited information in the training set will wear too thin, and the statistical test will show that there isn’t enough data to confirm the connection is real. A bigger training set might confirm the hypothesis – or the putative idea might not stand up to scrutiny.

Execution

Once training is complete, the AI can use its chosen rules to categorise the inputs – whether a bridge over a river, a good chess position, or a good stock to invest in. The bigger the training set, the better the outcome. If something is in fact undecidable from the input data, it will tell you that too. And, once learning is complete, these rules are almost instant to make use of.
As with Heisenberg’s uncertainty principle, there is a limit to what the AI can know. But with the ability to accumulate vast knowledge over time and space, this limit is far higher than any one person could attain.
Stephen B Streater
Founder and Director of R&D

The post AI – The limits of IQ appeared first on Blackbird.

But flying a plane is harder. Learning to fly a plane safely and getting a pilot’s licence is a long drawn out process. Maybe 10% could fly a plane. Building one is harder again – only 1% of people could actually make a plane. But then, what do they make? They need a design. 0.1% of people could design a plane. And who invented the core technology which these people design into their planes? Maybe the 0.01%.
When you are in the top 1%, you see and meet the 0.1% and the 0.01%, and realise the top 1% is not that special.
When I joined Mensa (the top 1%) as a student at Cambridge, all my friends got in easily – even the thick ones. Quite a few, like me, got the top IQ the test could give. Most of these were far more intelligent than me.
And being in the top 1% for wealth doesn’t feel that rich, living in Wimbledon Village. It’s probably average for my peers.
So, the thing about the top 1%, is that it’s fairly routine. When we think of great composers, artists or scientists of today, we are talking about the top 0.001%. And when we are talking about the greatest of all time, we are looking at the top 0.0001% or 0.00001%.

Non-computable problems

There are some problems which cannot be solved by any algorithm. This is not merely lack of imagination or programming ability; in many cases, you can prove that no algorithm can exist. Alan Turing demonstrated this with his proof that the halting problem could not be solved on his theoretical computer design, despite its unlimited memory.
Some people have conjectured that human brains are non-algorithmic, and computers will never catch up. But this is just superstition. There is no connection demonstrated between the limits of algorithms and the processes needed for intelligence. The rapid progress in AI has shown in an ever increasing range of tasks how limited human brains really are.

Getting AI into the top 1%

Despite this technical handicap, AI’s have a few tricks which will help them achieve greatness.
When people invented writing, we could suddenly benefit from knowledge discovered (and created) by a wide range of people over an extended period of time. A few hundred years ago, a polymath could still expect to know “everything”. But now, you can spend your whole life learning and understanding an obscure branch of knowledge – algebraic geometry, for example.
But with computers we have an advantage: their knowledge is cumulative. A computer can know everything (useful) that has ever been published, without having to spend a big chunk of its life learning it first.

Invent vs verify

Just as with designing a plane and flying on one, there is a big difference in difficulty between finding a solution and checking a solution: finding solutions is much harder.
For an AI, as for a human, understanding comes slowly. Learning is an arduous process. Invention – spotting new connections within the training data – is even slower. In my experiments, invention time grows faster than exponentially as a function of problem complexity. There are just so many possibilities to explore and test. But, for a worthwhile problem, an AI has the time and patience to invent.
Execution – acting on an existing understanding, on the other hand, is relatively quick – and easily replicable on the modern internet.
An AI with “unlimited” intelligence cannot be far away now. I will discuss my experiments in this area in a later post.
Advances in chess, Go – and the potential of an unlimited AI – have shown that reaching the 1% is merely a stepping stone for AI.
Stephen B Streater
Founder and Director of R&D

The post AI – The 1% appeared first on Blackbird.

Mind games provide a rich seam for AI. Every year, AI gains more scalps – recently including Go and Poker.
These games share a common feature: they are well defined – you can write down the rules for Go. Developers can solve this narrow problem using specialist techniques.

Real world problems

In the real world, many interesting problems are more open: make ethical investments; write a video codec; drive a car; educate our next generation; negotiate Brexit; make a General AI.
How do we even explain the problem to the AI? The problem starts with defining the problem.
This is normally achieved by using lots of training data – and getting the AI to use inductive logic (if you are doing it properly), or more likely, evolving a neural network, choosing variations which seems to get it about right most of the time as far as you can see.
When it is a program playing a board game, this is fine. When it is a self-driving car, this is risky. Though to be fair, self driving cars are safer than human drivers. The risk is that something unexpected happens – like unusual weather – and millions of cars crash.

Migrating AI into the real world

The pressure for AI is unstoppable, barring concerns about the Fermi Paradox.
Despite popular fiction – and chess programs – not every AI has to be as good as the best person in the world. Assistants such as spelling and grammar checkers, web purchase recommendations and translators give a natural route in. Even steady real world progress adds up to a transformation over a generation.
But anything that makes money directly will lead to a rapid virtuous circle through reinvestment.

Low hanging fruit

Making value is harder than spotting value.
My Granny cynically explained that to sell you need to appeal to three emotions: greed, fear and vanity. Financial markets are simpler – greed and fear is enough. Buy when there is blood on the streets.
When I wrote my AI, for real time image recognition, I planned to use the AI engine for share trading. In the meantime, I made a simple spreadsheet to tide me over. But it turned out that even very simple tactics beat the market so convincingly I didn’t even draw a salary for my first 14 years a Forbidden – until the institutions said it messed up their cost model KPIs.
Disappointingly though, the spreadsheet made money off other traders. Maybe it provided some valuable liquidity in the market as a side effect, but it basically moved wealth from other people to me without actually make any new wealth.
Having said that, my best investments have been long term holdings in tech companies where I’ve got in years before the market has recognised value. In investment, you make your money by choosing well – and waiting. This is where AIs will ultimately win out.
The Paradise Papers of the future will be reporting that the offshore holders are computer programs running in the cloud.
Stephen B Streater
Founder and Director of R&D

The post AI – the Real World appeared first on Blackbird.

Can perspiration result in inspiration?

As it happens, we have a good way of testing the effectiveness of brute force processing – the game of chess.
For years, chess was the holy grail of AI – the ultimate proof of human superiority. As supreme chess champion, Garry Kasparov was a natural target for AI researchers. And in 1997, the relentless advances in computer power finally enabled an upgraded Deep Blue to beat Garry Kasparov.
Luckily, I already had my own chess program.

Computer chess techniques

Computer chess programs have a position evaluation function. This is an estimate of how good a position is. This approximation is enhanced by a depth first search.
Computer chess programs look at possible moves (a 1-ply search) and possible replies to those moves (a 2-ply search) – and possible replies to those moves (a 3-ply search) – up to the search depth.
Ending the search is tricky because of the horizon problem: who gets the last word? It is tempting to make a bold capture on the last move in the search, as there is no depth left for the withering response! A quiescence search is a slimmed down search which ends at a stable position.
The possible moves at each ply give rise to a search tree – which is chess has a branching factor of around 40.
Looking at this tree, the minimax algorithm looks for the principal variation – the best moves and responses for each player.
A deeper search leads to better play, but the tree grows exponentially. Even at two ply, you are looking at 40×40=1600 positions, and a ten ply search gives around 10000000000000000 positions.
The remarkable alpha-beta algorithm gives a move with the same score as minimax, while reducing the number of positions evaluated to (if you are lucky) the square root of the original number. 10000000000000000 positions could become 100000000 positions – billions of times faster.
Alpha-beta works effectively when looking at the best moves first so sorting the moves is important. As shallow searches are so much faster than deeper ones, sequential searches at increasing depth – iterative deepening – helps in ordering moves.
It turns out the search tree is more of a thicket: transposing moves can often get you to the same position. You can lop off entire branches if you recognise a position you have been in before. Storing positions in a hash table, allowing quick access, gives a huge improvement.

Optimisations

To understand the effect of depth alone, my algorithm delivered an exhaustive search at each depth. On an early home computer, to get to an exhaustive 12 ply search, I had to invoke many algorithmic optimisations and extreme programming.
With the techniques listed above, an ultra simple evaluation function with alpha-beta windowing gave an exhaustive search with a branching factor typically around 4.
Optimised ARM machine code which tested moves in just over 1 CPU instruction each and generated moves in 5-10 CPU instructions allowed it to crunch millions of positions.
While the so-called God’s algorithm is currently unknown in most chess positions, I was now ready to see what happens when you increase exhaustive search depth.

Results

At 3 ply, the computer was easy to beat, but it was very unforgiving of simple mistakes, snaffling loose pawns or pieces off beginners at every opportunity. This was a early sign of the immense power of exhaustive search.
I could now crank up the IQ in stages – by increasing the search depth. I felt like the Time Traveller compelled to move into an unknown future.
Changing up a gear to 4 ply ruled out my simple tricks. At 5 ply it was almost as good as me. Each ply was making a noticeable difference in its ability.
6 ply beat me most of the time. In the First International Computer Games Olympiad, it repeatedly survived against better but slower programs, like James Bond at his best, through a series of implausible escapes.
With improvements to the code – and the StrongARM CPU – the program was crunching through up to 50 million moves per second, making even 12 ply easily testable, and 18 ply taking less than a second per move in the end game.
Its discipline improved my play enough to equal it at 8 ply, but above that it seemed indestructible.
And now we can see how the program got better with each extra ply: by looking deeper, an understanding was emerging. At 4 ply, it could not only follow every one of its pieces for two moves, it could look at every combination of two pieces working together (and two of the opponent’s pieces likewise). At 6 ply, it was coordinating three pieces in an attacking (or defensive) move. By twelve ply it has six pieces working together. From mere moves a plan emerges.
This ability to combine multiple ideas into a bigger concepts is key to intelligence.
So now it was time to bring in a real chess player – my father. A mathematician, the youngest (full) Professor at London University in the 20th century, he played chess internationally when a student. By his hundredth international tournament game in Switzerland, he could remember every move in every game. He used to play my school friends blindfold four at a time, and win every game.
Well, he played my program at all the plies up to 12. His comment was: they are chess moves, but it isn’t chess. He easily won all the games.

Conclusions

My program’s exhaustive search was reinventing its strategy every move. Like a child who hasn’t practised, it was sight reading at every piano lesson. It had speed but not the crucial element: experience.
AI needs the ability to compress knowledge into a simple form for use in a more sophisticated search: it needs the ability to learn.
My chess experiments gave a glimpse of an unlimited AI. Each extra ply of search gives an improvement in conceptual depth. Intelligence is not a on-off switch. Intelligence is an emergent property.
Unusually designed as it was, my chess program was good enough to beat almost everyone on Earth at chess.
And so yes: perspiration does indeed lead to inspiration.
Stephen B Streater
Founder and Director of R&D

The post AI – perspiration or inspiration? appeared first on Blackbird.