Facebook(s fb) has published a research paper explaining a project called DeepFace that’s almost as good at putting names to faces as humans are. In fact, it might be better. The company claims its system, which is built using deep neural network, or deep learning, techniques, performed with 97.25 percent accuracy on a dataset commonly used to measure the effectiveness of facial recognition systems.
MIT Technology Review first reported on the DeepFace paper, which Facebook researchers are presenting at the IEEE Conference on Computer Vision and Pattern Recognition in June.
Deep learning is currently an area of investment for a number of web companies ranging from Pinterest to Netflix(s nflx), although Facebook and Google(s goog) have probably made the biggest news with their high-profile hires and acquisitions. It’s such a hot field because deep learning techniques are proving very effective at recognizing objects within images and analyzing…
Ver o post original 390 mais palavras
Hadoop software vendor Cloudera didn’t make a lot of waves when it bought a London-based startup called Myrrix last year, and it hasn’t made a lot of noise about the company’s machine learning technology since then. But the company’s technology and its founder, Sean Owen, could turn out to be very valuable assets.
Owen, whose official title is director of data science, now spends him time working on an open source machine learning project called Oryx. (It’s a species of African antelope; Cloudera also sells a product called Impala). Oryx is intended to help Hadoop users build machine learning models and then deploy them so they can be queried and serve results in real time, say as part of a spam filter or a recommendation engine. Ideally, Oryx will also suuport models that can update themselves as data streams in.
Owen calls it the difference between Hadoop’s traditional sweet spot…
Ver o post original 641 mais palavras
- Início: 17 Fev 2014
- Duração do curso: 10 semanas
- Esforço estimado: 6h-8h por semana
Depois de tanto tempo, a terceira parte da série Bias vs. Variância saiu!
Apenas relembrando, a série foi dividida da seguinte forma:
- Primeira parte: conceito de bias e variância
- Segunda parte: cálculo do bias e da variância
- Terceira parte: métodos de redução do bias e da variância
Em resumo, enquanto o bias está ligado à capacidade das predições do modelo se aproximarem dos valores reais, a variância está relacionada à consistência dos resultados do modelo em diferentes conjuntos de dados.
Apesar de sabermos calcular “explicitamente” o bias e a variância de um modelo, dependendo do tamanho do conjunto de dados (número de variáveis e samples) e da complexidade do modelo, o processo pode ser computacionalmente caro. Desta forma, precisamos ter outras maneiras de inferir se estamos com problemas de bias ou variância.
Say a web publisher wants to find out which banner ad is most appealing to which audience, or which price point will make a certain user more likely to buy. Normally it would use multivariate A/B testing — the process of showing different versions of the same screen or screen elements to users and gathering data on their reactions — but the process is lengthy and testing numerous variables like location, time of day, or browser used spreads the data thin.
The Ireland-based operation uses A/B testing, machine learning and basic user data garnered from IP addresses and user agent. As the API receives user feedback — did she click on a banner or not? — Synference detects patterns of user behavior and updates its statistical model accordingly. It also allows companies to exploit this information before…
Ver o post original 187 mais palavras
A peek at the early days of the Quantum AI Lab: a partnership between NASA, Google, and a 512-qubit D-Wave Two quantum computer.
Google (s goog) silently did something revolutionary on Thursday. It open sourced a tool called word2vec, prepackaged deep-learning software designed to understand the relationships between words with no human guidance. Just input a textual data set and let underlying predictive models get to work learning.
“This is a really, really, really big deal,” said Jeremy Howard, president and chief scientist of data-science competition platform Kaggle. “… It’s going to enable whole new classes of products that have never existed before.” Think of Siri on steroids, for starters, or perhaps emulators that could mimic your writing style down to the tone.
When deep learning works, it works great
To understand Howard’s excitement, let’s go back a few days. It was Monday and I was watching him give a presentation in Chicago about how deep learning was dominating the competition in Kaggle, the online platform where organization present vexing predictive problems…
Ver o post original 1.323 mais palavras