-
Directional Deduction
Authors:
Roderick Batchelor
Abstract:
We present some new methods for logical deduction, based on ideas from ground theory. Roughly speaking, in our calculi a typical deduction will proceed as follows: we first analyse the premiss down to its ultimate grounds; then we discard information irrelevant to the conclusion; and then we synthesize the conclusion up from its ultimate grounds. We give a series of calculi for: classical proposit…
▽ More
We present some new methods for logical deduction, based on ideas from ground theory. Roughly speaking, in our calculi a typical deduction will proceed as follows: we first analyse the premiss down to its ultimate grounds; then we discard information irrelevant to the conclusion; and then we synthesize the conclusion up from its ultimate grounds. We give a series of calculi for: classical propositional logic (Chapter 1); classical predicate logic (Chapter 2); modal propositional logic and modal predicate logic (Chapter 3); and some relevantistic fragments of these various systems (Chapters 4 and 5). In connection with these fragments we develop also some new semantic constructions, of 'truthmaker semantics' type.
△ Less
Submitted 8 August, 2022;
originally announced August 2022.
-
Continual Learning Augmented Investment Decisions
Authors:
Daniel Philps,
Tillman Weyde,
Artur d'Avila Garcez,
Roy Batchelor
Abstract:
Investment decisions can benefit from incorporating an accumulated knowledge of the past to drive future decision making. We introduce Continual Learning Augmentation (CLA) which is based on an explicit memory structure and a feed forward neural network (FFNN) base model and used to drive long term financial investment decisions. We demonstrate that our approach improves accuracy in investment dec…
▽ More
Investment decisions can benefit from incorporating an accumulated knowledge of the past to drive future decision making. We introduce Continual Learning Augmentation (CLA) which is based on an explicit memory structure and a feed forward neural network (FFNN) base model and used to drive long term financial investment decisions. We demonstrate that our approach improves accuracy in investment decision making while memory is addressed in an explainable way. Our approach introduces novel remember cues, consisting of empirically learned change points in the absolute error series of the FFNN. Memory recall is also novel, with contextual similarity assessed over time by sampling distances using dynamic time war** (DTW). We demonstrate the benefits of our approach by using it in an expected return forecasting task to drive investment decisions. In an investment simulation in a broad international equity universe between 2003-2017, our approach significantly outperforms FFNN base models. We also illustrate how CLA's memory addressing works in practice, using a worked example to demonstrate the explainability of our approach.
△ Less
Submitted 25 January, 2019; v1 submitted 5 December, 2018;
originally announced December 2018.
-
X-ray absorption spectroscopy on layered cobaltates Na_xCoO_2
Authors:
T. Kroll,
M. Knupfer,
J. Geck,
C. Hess,
T. Schwieger,
G. Krabbes,
C. Sekar,
D. R. Batchelor,
H. Berger,
B. Buechner
Abstract:
Measurements of polarization and temperature dependent soft x-ray absorption have been performed on Na_xCoO_2 single crystals with x=0.4 and x=0.6. They show a deviation of the local trigonal symmetry of the CoO_6 octahedra, which is temperature independent in a temperature range between 25 K and 372 K. This deviation was found to be different for Co^{3+} and Co^{4+} sites. With the help of a cl…
▽ More
Measurements of polarization and temperature dependent soft x-ray absorption have been performed on Na_xCoO_2 single crystals with x=0.4 and x=0.6. They show a deviation of the local trigonal symmetry of the CoO_6 octahedra, which is temperature independent in a temperature range between 25 K and 372 K. This deviation was found to be different for Co^{3+} and Co^{4+} sites. With the help of a cluster calculation we are able to interpret the Co L_{23}-edge absorption spectrum and find a do** dependent energy splitting between the t_{2g} and the e_g levels (10Dq) in Na_xCoO_2.
△ Less
Submitted 20 June, 2006;
originally announced June 2006.
