Anamorphic Stretch Transform: Analog and Digital Compression of Big Data

Telecommunication systems routinely generate, capture and analyze data at rates exceeding billions of bits per second.  Interestingly, the scale of the problem is similar to that of blood analysis. With approximately 1 billion cells per milliliter of blood, detection of a few abnormal cells in a blood sample translates into a "cell error rate" of 10-12, a value that is curiously similar to the bit error rate in telecommunication systems.

Motivated by WDM and the time-stretch technique, a new type of bright-field camera, known as STEAM, has demonstrated imaging of cells with record shutter speed and throughput leading to detection of rare breast cancer cells in blood with one-in-a-million sensitivity. A second technique called FIRE is a new approach to fluorescent imaging that is based on wireless communication techniques. FIRE has achieved real-time pixel readout rates one order of magnitude faster than the current gold standard in high-speed fluorescence imaging.   Producing data rates as high as one Terabit per second, these real-time instruments pose a big data challenge that overwhelms even the most advanced computers. Driven by the necessity of solving this problem, we have recently introduced and demonstrated a categorically new data compression technology. The so called Anamorphic Stretch Transform is a physics-based compression technique that not only performs real-time data compression in the analog domain, but can also function as a digital algorithm for compression of images with superior performance to JPEG and other compression technologies. It alleviates the big data problem in real-time instruments, in digital imaging and beyond.

Speaker: Bahram Jalali, UCLA