A Sober Look at Spectral Learning
نویسندگان
چکیده
Spectral learning recently generated lots of excitement in machine learning, largely because it is the first known method to produce consistent estimates (under suitable conditions) for several latent variable models. In contrast, maximum likelihood estimates may get trapped in local optima due to the non-convex nature of the likelihood function of latent variable models. In this paper, we do an empirical evaluation of spectral learning (SL) and expectation maximization (EM), which reveals an important gap between the theory and the practice. First, SL often leads to negative probabilities. Second, EM often yields better estimates than spectral learning and it does not seem to get stuck in local optima. We discuss how the rank of the model parameters and the amount of training data can yield negative probabilities. We also question the common belief that maximum likelihood estimators are necessarily inconsistent.
منابع مشابه
Big Data and Learning Analytics in Blended Learning Environments: Benefits and Concerns
— The purpose of this article is to examine big data and learning analytics in blended learning environments. It will examine the nature of these concepts, provide basic definitions, and identify the benefits and concerns that apply to their development and implementation. This article draws on concepts associated with data-driven decision making, which evolved in the 1980s and 1990s, and takes...
متن کاملTwo classes of locally compact sober spaces
We deal with two classes of locally compact sober spaces, namely, the class of locally spectral coherent spaces and the class of spaces in which every point has a closed spectral neighborhood (CSN-spaces, for short). We prove that locally spectral coherent spaces are precisely the coherent sober spaces with a basis of compact open sets. We also prove that CSN-spaces are exactly the locally spec...
متن کاملEEG spectral power alternation & learning disability in rats exposure to lead through prenatal developing age
Introduction: Lead intoxication in children has been associated with neurodevelopment disabilities that may result in motor and cognitive impairment. The aim of the present study is to evaluate in long term effects of prenatal exposure to low-level Pb2+ on learning, memory and EEG. Materials & Methods: Rat pregnant were exposed to distilled water or 0.05, 0.1, 0.2 % of Pb-acetate in the dr...
متن کاملA Sober Look at Clustering Stability
Stability is a common tool to verify the validity of sample based algorithms. In clustering it is widely used to tune the parameters of the algorithm, such as the number k of clusters. In spite of the popularity of stability in practical applications, there has been very little theoretical analysis of this notion. In this paper we provide a formal definition of stability and analyze some of its...
متن کاملI'm No Longer a Child: A Closer Look at the Interaction Between Iranian EFL University Students' Identities and Their Academic Performance
Although university EFL students represent a wide array of social and cultural identities, their multiple and diverse identities are not usually considered in foreign language classrooms. This qualitative case study attempted to examine identity conflicts experienced by Iranian EFL learners at the university context. To this end, two Shiraz University students' identities were investigated. Sem...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- CoRR
دوره abs/1406.4631 شماره
صفحات -
تاریخ انتشار 2014