Learning A Physical Long-term Predictor

نویسندگان

  • Sébastien Ehrhardt
  • Aron Monszpart
  • Niloy J. Mitra
  • Andrea Vedaldi
چکیده

Evolution has resulted in highly developed abilities in many natural intelligences to quickly and accurately predict mechanical phenomena. Humans have successfully developed laws of physics to abstract and model such mechanical phenomena. In the context of artificial intelligence, a recent line of work has focused on estimating physical parameters based on sensory data and use them in physical simulators to make long-term predictions. In contrast, we investigate the effectiveness of a single neural network for end-to-end long-term prediction of mechanical phenomena. Based on extensive evaluation, we demonstrate that such networks can outperform alternate approaches having even access to ground-truth physical simulators, especially when some physical parameters are unobserved or not known a-priori. Further, our network outputs a distribution of outcomes to capture the inherent uncertainty in the data. Our approach demonstrates for the first time the possibility of making actionable long-term predictions from sensor data without requiring to explicitly model the underlying physical laws.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

P3: Mechanisms of TrkB-Mediated Hippocampal Long-Term Potentiation in Learning and Memory

Long-term potentiation (LTP) is a process that certain types of synaptic stimulation lead to a long-lasting enhancement in the strength of synaptic transmission. Studies in recent years indicate the importance of molecular pathways in the development of memory and learning. Tropomyosin receptor kinase B (TrkB) is a member of the neurotrophin receptor tyrosine kinase family, that its ligand is b...

متن کامل

The Effect of Mnemonic Key Word Method on Vocabulary Learning and Long Term Retention

Most of the studies on the key word method of second/foreign language vocabulary learning have been based on the evidence from laboratory experiments and have primarily involved the use of English key words to learn the vocabularies of other languages. Furthermore, comparatively quite limited number of such studies is done in authentic classroom contexts. The present study inquired into the eff...

متن کامل

The effect of intrahippocampal microinjection of Naloxone on short –term and long-term memory in adult male rats

Introduction:The hippocampus is one for the major centers of learning and memory. Role of the opioid system has been investigated and on the other hand receptors related to this system such as mu-opioid receptors (MOR) are extended in the hippocampus. In this study the effect of Naloxone administration as a mu opioid receptor antagonist on passive avoidance memory in adult male rats was i...

متن کامل

The Influence of Short-term and Long-term Memory on the Identification and Discrimination of Non-native Speech Sounds

This study examined two possible sources of individual differences in crosslanguage speech perception, the capacity to phonologically encode speech and short-term memory span. Phonological coding was defined as the ability to encode non-native contrasts as distinct phonemes based on representations in long-term memory. Short-term memory was defined as a fixed capacity regulating the extent of e...

متن کامل

Responses of macaque perirhinal neurons during and after visual stimulus association learning.

Recent lesion studies have implicated the perirhinal cortex in learning that two objects are associated, i.e., visual association learning. In this experiment we tested whether neuronal responses to associated stimuli in perirhinal cortex are altered over the course of learning. Neurons were recorded from monkeys during performance of a visual discrimination task in which a predictor stimulus w...

متن کامل

ذخیره 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 نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/1703.00247  شماره 

صفحات  -

تاریخ انتشار 2017