The last inequality follows Eq. (18).

Comcast is spinning out Rotten Tomatoes, Fandango and a bunch of NBCUniversal (NBCU) cable networks into a separate company. That means USA Network, CNBC, MSNBC, Oxygen, E!, SYFY and Golf Channel will soon have a new home. Comcast is hanging onto other NBCU operations, namely NBC, Peacock, film and TV studios, Telemundo and theme parks. Bravo is also sticking around to help keep feeding Peacock's ever-hungry reality TV maw. Comcast says the new entity will be a "tax-free spin-off" and the step is "expected to be accretive to revenue growth at Comcast and approximately neutral to Comcast's leverage position."

Behavioral cloning (BC) can recover a good policy from abundant expert data, but may fail when expert data is insufficient. This paper considers a situation where, besides the small amount of expert data, a supplementary dataset is available, which can be collected cheaply from sub-optimal policies. Imitation learning with a supplementary dataset is an emergent practical framework, but its theoretical foundation remains under-developed. To advance understanding, we first investigate a direct extension of BC, called NBCU, that learns from the union of all available data. Our analysis shows that, although NBCU suffers an imitation gap that is larger than BC in the worst case, there exist special cases where NBCU performs better than or equally well as BC. This discovery implies that noisy data can also be helpful if utilized elaborately. Therefore, we further introduce a discriminator-based importance sampling technique to re-weight the supplementary data, proposing the WBCU method. With our newly developed landscape-based analysis, we prove that WBCU can outperform BC in mild conditions. Empirical studies show that WBCU simultaneously achieves the best performance on two challenging tasks where prior state-of-the-art methods fail.

Revolutions, it's been remarked, never go backward. Nor do they advance at a constant rate. Consider the immense transformation unleashed by data analytics. By now, it's clear the data revolution is changing businesses and industries in profound and unalterable ways. But the changes are neither uniform nor linear, and companies' data-analytics efforts are all over the map. McKinsey research suggests that the gap between leaders and laggards in adopting analytics, within and among industry sectors, is growing. Some companies are doing amazing things; some are still struggling with the basics; and some are feeling downright overwhelmed, with executives and members of the rank and file questioning the return on data initiatives. For leading and lagging companies alike, the emergence of data analytics as an omnipresent reality of modern organizational life means that a healthy data culture is becoming increasingly important. With that in mind, we've spent the past few months talking with analytics leaders at companies from a wide range of industries and geographies, drilling down on the organizing principles, motivations, and approaches that undergird their data efforts. We're struck by themes that recur over and again, including the benefits of data, and the risks; the skepticism from employees before they buy in, and the excitement once they do; the need for flexibility, and the insistence on common frameworks and tools. And, especially: the competitive advantage unleashed by a culture that brings data talent, tools, and decision making together. The experience of these leaders, and our own, suggests that you can't import data culture and you can't impose it. Most of all, you can't segregate it.